Hybrid Arima - LSTM code - Sentiment¶
The hybrid ARIMA-LSTM model is open to a variety of experimentation. For ideal performance, a balance must be reached between the levels of volatility that work best for the ARIMA and LSTM models. Using shorter MA periods that result in a non-mesokurtic distribution may achieve a better volatility balance between models.
Import Libraries¶
In [17]:
import pandas as pd
pd.set_option('display.max_rows', 500)
import timeit
In [18]:
!pip install -q -U keras-tuner
|████████████████████████████████| 98 kB 3.2 MB/s
In [19]:
import keras_tuner as kt
In [20]:
!pip install pmdarima
Collecting pmdarima
Downloading pmdarima-1.8.4-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.manylinux_2_24_x86_64.whl (1.4 MB)
|████████████████████████████████| 1.4 MB 4.8 MB/s
Requirement already satisfied: scipy>=1.3.2 in /usr/local/lib/python3.7/dist-packages (from pmdarima) (1.4.1)
Requirement already satisfied: urllib3 in /usr/local/lib/python3.7/dist-packages (from pmdarima) (1.24.3)
Requirement already satisfied: joblib>=0.11 in /usr/local/lib/python3.7/dist-packages (from pmdarima) (1.1.0)
Requirement already satisfied: setuptools!=50.0.0,>=38.6.0 in /usr/local/lib/python3.7/dist-packages (from pmdarima) (57.4.0)
Requirement already satisfied: pandas>=0.19 in /usr/local/lib/python3.7/dist-packages (from pmdarima) (1.1.5)
Collecting statsmodels!=0.12.0,>=0.11
Downloading statsmodels-0.13.1-cp37-cp37m-manylinux_2_17_x86_64.manylinux2014_x86_64.whl (9.8 MB)
|████████████████████████████████| 9.8 MB 33.8 MB/s
Requirement already satisfied: numpy>=1.19.3 in /usr/local/lib/python3.7/dist-packages (from pmdarima) (1.19.5)
Requirement already satisfied: Cython!=0.29.18,>=0.29 in /usr/local/lib/python3.7/dist-packages (from pmdarima) (0.29.24)
Requirement already satisfied: scikit-learn>=0.22 in /usr/local/lib/python3.7/dist-packages (from pmdarima) (1.0.1)
Requirement already satisfied: python-dateutil>=2.7.3 in /usr/local/lib/python3.7/dist-packages (from pandas>=0.19->pmdarima) (2.8.2)
Requirement already satisfied: pytz>=2017.2 in /usr/local/lib/python3.7/dist-packages (from pandas>=0.19->pmdarima) (2018.9)
Requirement already satisfied: six>=1.5 in /usr/local/lib/python3.7/dist-packages (from python-dateutil>=2.7.3->pandas>=0.19->pmdarima) (1.15.0)
Requirement already satisfied: threadpoolctl>=2.0.0 in /usr/local/lib/python3.7/dist-packages (from scikit-learn>=0.22->pmdarima) (3.0.0)
Requirement already satisfied: patsy>=0.5.2 in /usr/local/lib/python3.7/dist-packages (from statsmodels!=0.12.0,>=0.11->pmdarima) (0.5.2)
Installing collected packages: statsmodels, pmdarima
Attempting uninstall: statsmodels
Found existing installation: statsmodels 0.10.2
Uninstalling statsmodels-0.10.2:
Successfully uninstalled statsmodels-0.10.2
Successfully installed pmdarima-1.8.4 statsmodels-0.13.1
In [21]:
import pmdarima
In [22]:
url = 'https://launchpad.net/~mario-mariomedina/+archive/ubuntu/talib/+files'
!wget $url/libta-lib0_0.4.0-oneiric1_amd64.deb -qO libta.deb
!wget $url/ta-lib0-dev_0.4.0-oneiric1_amd64.deb -qO ta.deb
!dpkg -i libta.deb ta.deb
!pip install ta-lib
import talib
Selecting previously unselected package libta-lib0.
(Reading database ... 155222 files and directories currently installed.)
Preparing to unpack libta.deb ...
Unpacking libta-lib0 (0.4.0-oneiric1) ...
Selecting previously unselected package ta-lib0-dev.
Preparing to unpack ta.deb ...
Unpacking ta-lib0-dev (0.4.0-oneiric1) ...
Setting up libta-lib0 (0.4.0-oneiric1) ...
Setting up ta-lib0-dev (0.4.0-oneiric1) ...
Processing triggers for man-db (2.8.3-2ubuntu0.1) ...
Processing triggers for libc-bin (2.27-3ubuntu1.3) ...
/sbin/ldconfig.real: /usr/local/lib/python3.7/dist-packages/ideep4py/lib/libmkldnn.so.0 is not a symbolic link
Collecting ta-lib
Downloading TA-Lib-0.4.22.tar.gz (268 kB)
|████████████████████████████████| 268 kB 4.3 MB/s
Installing build dependencies ... done
Getting requirements to build wheel ... done
Preparing wheel metadata ... done
Requirement already satisfied: numpy in /usr/local/lib/python3.7/dist-packages (from ta-lib) (1.19.5)
Building wheels for collected packages: ta-lib
Building wheel for ta-lib (PEP 517) ... done
Created wheel for ta-lib: filename=TA_Lib-0.4.22-cp37-cp37m-linux_x86_64.whl size=1465669 sha256=19ce1fdaaa80a58ecfab1c42d146c7b72f2231b92e162e833e2364e590e21be8
Stored in directory: /root/.cache/pip/wheels/7b/63/a9/144081748d9c4f0a09b4670c7b3c414bcb33ff97f0724c753a
Successfully built ta-lib
Installing collected packages: ta-lib
Successfully installed ta-lib-0.4.22
In [23]:
import tensorflow
import statsmodels.tsa.api
import keras
import sklearn
In [24]:
from tensorflow.keras.optimizers import Adam
from tensorflow.keras import Sequential
from tensorflow.keras.layers import Dense, LSTM, Dropout, Bidirectional,BatchNormalization, Embedding, TimeDistributed, LeakyReLU, GRU
from tensorflow.keras.callbacks import EarlyStopping, ModelCheckpoint, ReduceLROnPlateau
In [25]:
from keras.models import Sequential, load_model
from keras.layers import Dense, LSTM, Activation, Dropout
from keras import backend as K
from keras.utils.generic_utils import get_custom_objects
from keras.callbacks import ModelCheckpoint,EarlyStopping
from keras.regularizers import l1_l2
In [26]:
import math
In [27]:
from statsmodels.tsa.api import VAR
from statsmodels.tsa.statespace.varmax import VARMAX,VARMAXResults
In [28]:
from sklearn.metrics import mean_squared_error, mean_absolute_percentage_error, mean_absolute_error
from sklearn.preprocessing import MinMaxScaler
from sklearn.model_selection import train_test_split
In [29]:
from matplotlib import pyplot
In [30]:
import json
import datetime
import pandas as pd
import numpy as np
import os
from scipy.stats import kurtosis
import pmdarima as pm
from pmdarima import auto_arima
from talib import abstract
import json
import matplotlib.pyplot as plt
# plt.rcParams.update({'font.size': 16})
from matplotlib.pyplot import figure
from numpy import array
from numpy import hstack
from keras.models import Sequential
from keras.layers import LSTM
from keras.layers import Dense
from keras.layers import RepeatVector
from keras.layers import TimeDistributed
In [31]:
from keras.utils.generic_utils import get_custom_objects
from tensorflow.keras.utils import plot_model
In [32]:
import warnings
from statsmodels.tools.sm_exceptions import ConvergenceWarning
warnings.simplefilter('ignore', ConvergenceWarning)
Load Data¶
In [6]:
from google.colab import drive
drive.mount('/content/drive')
Drive already mounted at /content/drive; to attempt to forcibly remount, call drive.mount("/content/drive", force_remount=True).
In [40]:
cd drive/MyDrive/Stock price prediction/Generated datasets
/content/drive/.shortcut-targets-by-id/1IaGjVBlTspxI2CHSrxfYnaiYvsaG0pHs/Stock price prediction/Generated datasets
In [41]:
df = pd.read_csv("FULL_Data_google_COVID_bull_bear.csv",parse_dates=[0])
df.tail(10)
Out[41]:
| Unnamed: 0 | Unnamed: 0.1 | Unnamed: 0.1.1 | Unnamed: 0.1.1.1 | Open | High | Low | Close | Adj Close | Volume | MA7 | MA21 | MACD | 20SD | upper_band | lower_band | EMA | logmomentum | absolute of 3 comp | angle of 3 comp | absolute of 6 comp | angle of 6 comp | absolute of 9 comp | angle of 9 comp | Date | search | COVID positiveIncrease | COVID deathIncrease | bull score | bear score | fourier bull 10 | fourier bull 30 | fourier bear 10 | fourier bear 30 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1592 | 1592 | 1781 | 1781 | 1781 | 150.199997 | 151.429993 | 150.059998 | 150.809998 | 150.809998 | 56787900.0 | 150.565717 | 148.423811 | -1.137777 | 2.817933 | 154.059677 | 142.787944 | 150.767809 | 5.009368 | 93.428749 | -0.061228 | 100.779503 | -0.039111 | 103.599003 | -0.022436 | 2021-11-09 | 19 | 112313 | 1258 | 0.119141 | 0.111328 | NaN | NaN | NaN | NaN |
| 1593 | 1593 | 1782 | 1782 | 1782 | 150.020004 | 150.130005 | 147.850006 | 147.919998 | 147.919998 | 65187100.0 | 150.417145 | 148.729049 | -1.236913 | 2.144358 | 153.017766 | 144.440332 | 148.869268 | 4.989888 | 92.922909 | -0.061683 | 99.694365 | -0.039762 | 101.872301 | -0.022657 | 2021-11-10 | 19 | 80301 | 1470 | 0.154297 | 0.109375 | NaN | NaN | NaN | NaN |
| 1594 | 1594 | 1783 | 1783 | 1783 | 148.960007 | 149.429993 | 147.679993 | 147.869995 | 147.869995 | 41000000.0 | 150.110001 | 149.060477 | -1.165047 | 1.767475 | 152.595428 | 145.525526 | 148.203086 | 4.989548 | 92.416471 | -0.062129 | 98.604584 | -0.040391 | 100.137594 | -0.022839 | 2021-11-11 | 19 | 94975 | 1662 | 0.102845 | 0.126915 | NaN | NaN | NaN | NaN |
| 1595 | 1595 | 1784 | 1784 | 1784 | 148.429993 | 150.399994 | 147.479996 | 149.990005 | 149.990005 | 63632600.0 | 149.895715 | 149.357144 | -0.869308 | 1.420732 | 152.198608 | 146.515681 | 149.394365 | 5.003879 | 91.909483 | -0.062566 | 97.510555 | -0.040998 | 98.396260 | -0.022980 | 2021-11-12 | 19 | 55499 | 797 | 0.157277 | 0.080595 | NaN | NaN | NaN | NaN |
| 1596 | 1596 | 1785 | 1785 | 1785 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 2021-11-13 | 19 | 146529 | 2505 | 0.139459 | 0.083243 | NaN | NaN | NaN | NaN |
| 1597 | 1597 | 1786 | 1786 | 1786 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 2021-11-14 | 19 | 40964 | 479 | 0.151261 | 0.100840 | NaN | NaN | NaN | NaN |
| 1598 | 1598 | 1787 | 1787 | 1787 | 150.369995 | 151.880005 | 149.429993 | 150.000000 | 150.000000 | 59222800.0 | 149.758571 | 149.602859 | -0.907641 | 1.229694 | 152.062246 | 147.143471 | 149.798122 | 5.003946 | 91.401994 | -0.062993 | 96.412672 | -0.041581 | 96.649685 | -0.023077 | 2021-11-15 | 22 | 30290 | 148 | 0.136737 | 0.109389 | NaN | NaN | NaN | NaN |
| 1599 | 1599 | 1788 | 1788 | 1788 | 149.940002 | 151.490005 | 149.339996 | 151.000000 | 151.000000 | 59256200.0 | 149.718571 | 149.814763 | -0.791320 | 1.236243 | 152.287250 | 147.342277 | 150.599374 | 5.010635 | 90.894052 | -0.063410 | 95.311334 | -0.042140 | 94.899260 | -0.023130 | 2021-11-16 | 22 | 138962 | 1294 | 0.135531 | 0.115385 | NaN | NaN | NaN | NaN |
| 1600 | 1600 | 1789 | 1789 | 1789 | 151.000000 | 155.000000 | 150.990005 | 153.490005 | 153.490005 | 88807000.0 | 150.154286 | 150.040002 | -0.657719 | 1.467121 | 152.974245 | 147.105759 | 152.526461 | 5.027099 | 90.385704 | -0.063817 | 94.206941 | -0.042673 | 93.146378 | -0.023135 | 2021-11-17 | 22 | 87626 | 1290 | 0.100870 | 0.126957 | NaN | NaN | NaN | NaN |
| 1601 | 1601 | 1790 | 1790 | 1790 | 153.710007 | 158.669998 | 153.050003 | 157.869995 | 157.869995 | 137659100.0 | 151.162857 | 150.450002 | -0.609656 | 2.267825 | 154.985653 | 145.914351 | 156.088817 | 5.055417 | 89.877000 | -0.064214 | 93.099895 | -0.043179 | 91.392433 | -0.023090 | 2021-11-18 | 22 | 111404 | 1637 | 0.145098 | 0.121569 | NaN | NaN | NaN | NaN |
In [ ]:
cd ..
In [45]:
cd Archana - LSTM Hybrid/Outputs/sentiment
/content/drive/.shortcut-targets-by-id/1IaGjVBlTspxI2CHSrxfYnaiYvsaG0pHs/Stock price prediction/Archana - LSTM Hybrid/Outputs/sentiment
In [47]:
pd.to_datetime(df[np.isnan(df.Close)==True]['Date']).dt.day_name().head(5)
Out[47]:
0 Saturday 1 Sunday 3 Tuesday 7 Saturday 8 Sunday Name: Date, dtype: object
In [46]:
len(pd.to_datetime(df[np.isnan(df.Close)==True]['Date']).dt.day_name())
Out[46]:
497
In [48]:
len(df)
Out[48]:
1602
In [49]:
len(df) - len(pd.to_datetime(df[np.isnan(df.Close)==True]['Date']).dt.day_name())
Out[49]:
1105
In [32]:
df.dropna(inplace=True)
len(df)
Out[32]:
1080
In [51]:
pd.to_datetime(df[np.isnan(df.Close)==True]['Date']).dt.day_name().head(5)
Out[51]:
0 Saturday 1 Sunday 3 Tuesday 7 Saturday 8 Sunday Name: Date, dtype: object
In [52]:
df.head(5)
Out[52]:
| Unnamed: 0 | Unnamed: 0.1 | Unnamed: 0.1.1 | Unnamed: 0.1.1.1 | Open | High | Low | Close | Adj Close | Volume | MA7 | MA21 | MACD | 20SD | upper_band | lower_band | EMA | logmomentum | absolute of 3 comp | angle of 3 comp | absolute of 6 comp | angle of 6 comp | absolute of 9 comp | angle of 9 comp | Date | search | COVID positiveIncrease | COVID deathIncrease | bull score | bear score | fourier bull 10 | fourier bull 30 | fourier bear 10 | fourier bear 30 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 0 | 189 | 189 | 189 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 2017-07-01 | 17 | 0 | 0 | 0.000000 | 0.00 | 0.141086 | 0.147308 | 0.100437 | 0.101678 |
| 1 | 1 | 190 | 190 | 190 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 2017-07-02 | 17 | 0 | 0 | 0.200000 | 0.20 | 0.141930 | 0.147118 | 0.100488 | 0.100526 |
| 2 | 2 | 191 | 191 | 191 | 36.220001 | 36.325001 | 35.775002 | 35.875000 | 34.054882 | 57111200.0 | 36.173571 | 36.751904 | 0.303356 | 0.960520 | 38.672945 | 34.830864 | 35.924548 | 3.551770 | 38.458011 | 0.046984 | 29.704545 | 0.102857 | 43.304973 | -0.053955 | 2017-07-03 | 15 | 0 | 0 | 0.666667 | 0.00 | 0.142778 | 0.146810 | 0.100537 | 0.099251 |
| 3 | 3 | 192 | 192 | 192 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 2017-07-04 | 15 | 0 | 0 | 0.000000 | 0.25 | 0.143631 | 0.146382 | 0.100585 | 0.097860 |
| 4 | 4 | 193 | 193 | 193 | 35.922501 | 36.197498 | 35.680000 | 36.022499 | 34.194897 | 86278400.0 | 36.095357 | 36.634762 | 0.328795 | 0.852735 | 38.340231 | 34.929292 | 35.989849 | 3.555991 | 38.240991 | 0.049445 | 29.954520 | 0.099254 | 43.438321 | -0.053936 | 2017-07-05 | 15 | 0 | 0 | 0.400000 | 0.00 | 0.144487 | 0.145833 | 0.100630 | 0.096361 |
In [53]:
stock_col= list(df.columns)
stock_col = stock_col[4:len(stock_col)]
In [54]:
dataset_final = df[stock_col]
In [55]:
dataset_final.head(5)
Out[55]:
| Open | High | Low | Close | Adj Close | Volume | MA7 | MA21 | MACD | 20SD | upper_band | lower_band | EMA | logmomentum | absolute of 3 comp | angle of 3 comp | absolute of 6 comp | angle of 6 comp | absolute of 9 comp | angle of 9 comp | Date | search | COVID positiveIncrease | COVID deathIncrease | bull score | bear score | fourier bull 10 | fourier bull 30 | fourier bear 10 | fourier bear 30 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 2017-07-01 | 17 | 0 | 0 | 0.000000 | 0.00 | 0.141086 | 0.147308 | 0.100437 | 0.101678 |
| 1 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 2017-07-02 | 17 | 0 | 0 | 0.200000 | 0.20 | 0.141930 | 0.147118 | 0.100488 | 0.100526 |
| 2 | 36.220001 | 36.325001 | 35.775002 | 35.875000 | 34.054882 | 57111200.0 | 36.173571 | 36.751904 | 0.303356 | 0.960520 | 38.672945 | 34.830864 | 35.924548 | 3.551770 | 38.458011 | 0.046984 | 29.704545 | 0.102857 | 43.304973 | -0.053955 | 2017-07-03 | 15 | 0 | 0 | 0.666667 | 0.00 | 0.142778 | 0.146810 | 0.100537 | 0.099251 |
| 3 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 2017-07-04 | 15 | 0 | 0 | 0.000000 | 0.25 | 0.143631 | 0.146382 | 0.100585 | 0.097860 |
| 4 | 35.922501 | 36.197498 | 35.680000 | 36.022499 | 34.194897 | 86278400.0 | 36.095357 | 36.634762 | 0.328795 | 0.852735 | 38.340231 | 34.929292 | 35.989849 | 3.555991 | 38.240991 | 0.049445 | 29.954520 | 0.099254 | 43.438321 | -0.053936 | 2017-07-05 | 15 | 0 | 0 | 0.400000 | 0.00 | 0.144487 | 0.145833 | 0.100630 | 0.096361 |
Data Load for Experiments with Technical Indicators & Bull Bear¶
In [56]:
stock_col= list(df.columns)
stock_col1 = stock_col[4:len(stock_col)-9]
stock_col2 = stock_col[len(stock_col)-4:len(stock_col)]
for i in range(len(stock_col2)):
stock_col1.append(stock_col2[i])
dataset_final = df[stock_col1]
dataset_final.head(5)
Out[56]:
| Open | High | Low | Close | Adj Close | Volume | MA7 | MA21 | MACD | 20SD | upper_band | lower_band | EMA | logmomentum | absolute of 3 comp | angle of 3 comp | absolute of 6 comp | angle of 6 comp | absolute of 9 comp | angle of 9 comp | Date | fourier bull 10 | fourier bull 30 | fourier bear 10 | fourier bear 30 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 2017-07-01 | 0.141086 | 0.147308 | 0.100437 | 0.101678 |
| 1 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 2017-07-02 | 0.141930 | 0.147118 | 0.100488 | 0.100526 |
| 2 | 36.220001 | 36.325001 | 35.775002 | 35.875000 | 34.054882 | 57111200.0 | 36.173571 | 36.751904 | 0.303356 | 0.960520 | 38.672945 | 34.830864 | 35.924548 | 3.551770 | 38.458011 | 0.046984 | 29.704545 | 0.102857 | 43.304973 | -0.053955 | 2017-07-03 | 0.142778 | 0.146810 | 0.100537 | 0.099251 |
| 3 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 2017-07-04 | 0.143631 | 0.146382 | 0.100585 | 0.097860 |
| 4 | 35.922501 | 36.197498 | 35.680000 | 36.022499 | 34.194897 | 86278400.0 | 36.095357 | 36.634762 | 0.328795 | 0.852735 | 38.340231 | 34.929292 | 35.989849 | 3.555991 | 38.240991 | 0.049445 | 29.954520 | 0.099254 | 43.438321 | -0.053936 | 2017-07-05 | 0.144487 | 0.145833 | 0.100630 | 0.096361 |
In [57]:
# Set the date to datetime data
datetime_series = pd.to_datetime(dataset_final['Date'])
datetime_index = pd.DatetimeIndex(datetime_series.values)
dataset_final = dataset_final.set_index(datetime_index)
dataset_final = dataset_final.sort_values(by='Date')
dataset_final = dataset_final.drop(columns='Date')
dataset_final.head(5)
Out[57]:
| Open | High | Low | Close | Adj Close | Volume | MA7 | MA21 | MACD | 20SD | upper_band | lower_band | EMA | logmomentum | absolute of 3 comp | angle of 3 comp | absolute of 6 comp | angle of 6 comp | absolute of 9 comp | angle of 9 comp | fourier bull 10 | fourier bull 30 | fourier bear 10 | fourier bear 30 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2017-07-01 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 0.141086 | 0.147308 | 0.100437 | 0.101678 |
| 2017-07-02 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 0.141930 | 0.147118 | 0.100488 | 0.100526 |
| 2017-07-03 | 36.220001 | 36.325001 | 35.775002 | 35.875000 | 34.054882 | 57111200.0 | 36.173571 | 36.751904 | 0.303356 | 0.960520 | 38.672945 | 34.830864 | 35.924548 | 3.551770 | 38.458011 | 0.046984 | 29.704545 | 0.102857 | 43.304973 | -0.053955 | 0.142778 | 0.146810 | 0.100537 | 0.099251 |
| 2017-07-04 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 0.143631 | 0.146382 | 0.100585 | 0.097860 |
| 2017-07-05 | 35.922501 | 36.197498 | 35.680000 | 36.022499 | 34.194897 | 86278400.0 | 36.095357 | 36.634762 | 0.328795 | 0.852735 | 38.340231 | 34.929292 | 35.989849 | 3.555991 | 38.240991 | 0.049445 | 29.954520 | 0.099254 | 43.438321 | -0.053936 | 0.144487 | 0.145833 | 0.100630 | 0.096361 |
Train & test Dataset for Multistep Process¶
In [59]:
# Get features and target
X_value = pd.DataFrame(dataset_final.iloc[:, :])
y_value = pd.DataFrame(dataset_final.iloc[:, 3])
In [60]:
y_value.head(5)
Out[60]:
| Close | |
|---|---|
| 2017-07-01 | NaN |
| 2017-07-02 | NaN |
| 2017-07-03 | 35.875000 |
| 2017-07-04 | NaN |
| 2017-07-05 | 36.022499 |
In [61]:
# Normalized the data
X_scaler = MinMaxScaler(feature_range=(-1, 1))
y_scaler = MinMaxScaler(feature_range=(-1, 1))
X_scaler.fit(X_value)
y_scaler.fit(y_value)
Out[61]:
MinMaxScaler(feature_range=(-1, 1))
In [62]:
X_scale_dataset = X_scaler.fit_transform(X_value)
y_scale_dataset = y_scaler.fit_transform(y_value)
In [63]:
X_scale_dataset.shape, y_scale_dataset.shape,
Out[63]:
((1602, 24), (1602, 1))
In [64]:
X_value.shape[1]
Out[64]:
24
N Steps Definition¶
In [65]:
n_steps_in = 3
n_features = X_value.shape[1] #19 features
n_steps_out = 1
In [66]:
# Reshape the data
'''Set the data input steps and output steps,
we use 30 days data to predict 1 day price here,
reshape it to (None, input_step, number of features) used for LSTM input'''
# Get X/y dataset
def get_X_y(X_data, y_data):
X = list()
y = list()
yc = list()
length = len(X_data)
for i in range(0, length, 1):
# pdb.set_trace()
X_value = X_data[i: i + n_steps_in][:, :]
# print('[',i,': ',i,' + ',n_steps_in,'][:, :]')
y_value = y_data[i + n_steps_in: i + (n_steps_in + n_steps_out)][:, 0]
# print('[',i,' + ',n_steps_in,': ',i,' + (',n_steps_in,' + ',n_steps_out,')][:, 0]')
yc_value = y_data[i: i + n_steps_in][:, :]
if len(X_value) == 3 and len(y_value) == 1:
X.append(X_value)
y.append(y_value)
yc.append(yc_value)
return np.array(X), np.array(y), np.array(yc)
In [67]:
# get the train test predict index
def predict_index(dataset, X_train, n_steps_in, n_steps_out):
# get the predict data (remove the in_steps days)
train_predict_index = dataset.iloc[n_steps_in : X_train.shape[0] + n_steps_in + n_steps_out - 1, :].index
test_predict_index = dataset.iloc[X_train.shape[0] + n_steps_in:, :].index
return train_predict_index, test_predict_index
In [68]:
def mean_absolute_percentage_error(actual, prediction):
actual = pd.Series(actual)
prediction = pd.Series(prediction)
return 100 * np.mean(np.abs((actual - prediction))/actual)
In [69]:
# Split train/test dataset
def split_train_test(data):
train_size = round(len(X) * 0.75)
data_train = data[0:train_size]
data_test = data[train_size:]
return data_train, data_test
In [70]:
# Get data and check shape
X, y, yc = get_X_y(X_scale_dataset, y_scale_dataset)#X will be of shape 224 X 3 X 21 (each 3 X 21 array will be 3 days' worth of data). yc will have the corresponding closing price value
# pdb.set_trace()
X_train, X_test, = split_train_test(X)
y_train, y_test, = split_train_test(y)
yc_train, yc_test, = split_train_test(yc)
index_train, index_test, = predict_index(dataset_final, X_train, n_steps_in, n_steps_out)
In [71]:
# %% --------------------------------------- Save dataset -----------------------------------------------------------------
print('X shape: ', X.shape)
print('y shape: ', y.shape)
print('X_train shape: ', X_train.shape)
print('y_train shape: ', y_train.shape)
print('y_c_train shape: ', yc_train.shape)
print('X_test shape: ', X_test.shape)
print('y_test shape: ', y_test.shape)
print('y_c_test shape: ', yc_test.shape)
print('index_train shape:', index_train.shape)
print('index_test shape:', index_test.shape)
X shape: (1599, 3, 24) y shape: (1599, 1) X_train shape: (1199, 3, 24) y_train shape: (1199, 1) y_c_train shape: (1199, 3, 1) X_test shape: (400, 3, 24) y_test shape: (400, 1) y_c_test shape: (400, 3, 1) index_train shape: (1199,) index_test shape: (400,)
In [72]:
output_dim = y_train.shape[1]
output_dim
Out[72]:
1
In [73]:
df = dataset_final.copy()
In [74]:
df.rename(columns={'Date':'date','Open':'open','Low':'low','Close':'close','Volume':'volume','High':'high'}, inplace = True)
df.reset_index(drop=True,inplace=True)
In [75]:
df.head(1)
Out[75]:
| open | high | low | close | Adj Close | volume | MA7 | MA21 | MACD | 20SD | upper_band | lower_band | EMA | logmomentum | absolute of 3 comp | angle of 3 comp | absolute of 6 comp | angle of 6 comp | absolute of 9 comp | angle of 9 comp | fourier bull 10 | fourier bull 30 | fourier bear 10 | fourier bear 30 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 0.141086 | 0.147308 | 0.100437 | 0.101678 |
In [76]:
# df.drop(['volume', 'MACD','20SD','logmomentum','absolute of 3 comp','angle of 3 comp','absolute of 6 comp','angle of 6 comp','absolute of 9 comp','angle of 9 comp'], axis='columns', inplace=True) # only keep columns that can help as residuals in Arima Hybrid
In [77]:
df.head(1)
Out[77]:
| open | high | low | close | Adj Close | volume | MA7 | MA21 | MACD | 20SD | upper_band | lower_band | EMA | logmomentum | absolute of 3 comp | angle of 3 comp | absolute of 6 comp | angle of 6 comp | absolute of 9 comp | angle of 9 comp | fourier bull 10 | fourier bull 30 | fourier bear 10 | fourier bear 30 | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | 0.141086 | 0.147308 | 0.100437 | 0.101678 |
Train & Test Length¶
In [78]:
test_len = len(X_test)
In [79]:
train_len = len(X_train )
In [80]:
test_len, train_len
Out[80]:
(400, 1199)
Kurtosis Review¶
In [81]:
# Initialize moving averages from Ta-Lib, store functions in dictionary
# talib_moving_averages = ['SMA', 'EMA', 'WMA', 'DEMA', 'KAMA', 'MIDPOINT', 'MIDPRICE', 'T3', 'TEMA', 'TRIMA'] remove midprice due to outputbeing univariate
talib_moving_averages = ['SMA', 'EMA', 'WMA', 'DEMA', 'KAMA', 'MIDPOINT', 'T3', 'TEMA', 'TRIMA']
functions = {}
for ma in talib_moving_averages:
functions[ma] = abstract.Function(ma)
# Determine kurtosis "K" values for MA period 4-30
kurtosis_results = {'period': []}
for i in range(4, 100): # 100
kurtosis_results['period'].append(i)
for ma in talib_moving_averages:
# Run moving average, remove last N days (used later for test data set), trim MA result to last 30 days
ma_output = functions[ma](df[:-test_len], i).tail(14)
# Determine kurtosis "K" value
k = kurtosis(ma_output, fisher=False)
# add to dictionary
if ma not in kurtosis_results.keys():
kurtosis_results[ma] = []
kurtosis_results[ma].append(k)
kurtosis_results = pd.DataFrame(kurtosis_results)
kurtosis_results.to_csv('kurtosis_results.csv')
In [82]:
kurtosis_results.head(5)
Out[82]:
| period | SMA | EMA | WMA | DEMA | KAMA | MIDPOINT | T3 | TEMA | TRIMA | |
|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 4 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 1 | 5 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 2 | 6 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 3 | 7 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
| 4 | 8 | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN | NaN |
Optimized periods¶
In [83]:
# Determine period with K closest to 3 +/-5%
optimized_period = {}
# https://pypi.org/project/TA-Lib/ determines the type of moving average to use
# https://pandas.pydata.org/pandas-docs/version/0.17.0/generated/pandas.DataFrame.at.html#pandas.DataFrame.at
for ma in talib_moving_averages:
difference = np.abs(kurtosis_results[ma] - 3)
df_arimahyb = pd.DataFrame({'difference': difference, 'period': kurtosis_results['period']})
df_arimahyb = df_arimahyb.sort_values(by=['difference'], ascending=True).reset_index(drop=True)
if df_arimahyb.at[0, 'difference'] < 3 * 0.05:
optimized_period[ma] = df_arimahyb.at[0, 'period']
else:
print(ma + ' is not viable, best K greater or less than 3 +/-5%')
print('\nOptimized periods:', optimized_period)
SMA is not viable, best K greater or less than 3 +/-5%
EMA is not viable, best K greater or less than 3 +/-5%
WMA is not viable, best K greater or less than 3 +/-5%
DEMA is not viable, best K greater or less than 3 +/-5%
KAMA is not viable, best K greater or less than 3 +/-5%
MIDPOINT is not viable, best K greater or less than 3 +/-5%
T3 is not viable, best K greater or less than 3 +/-5%
TEMA is not viable, best K greater or less than 3 +/-5%
TRIMA is not viable, best K greater or less than 3 +/-5%
Optimized periods: {}
In [84]:
optimized_period
Out[84]:
{}
Simulation Keys¶
In [ ]:
simulation = {}
for ma in optimized_period:
print(ma)
print(functions[ma])
print ( int( optimized_period[ma]))
# if ma in ['EMA','WMA','DEMA','KAMA','MIDPOINT']:
# print(ma)
low_vol = df.apply(lambda c: functions[ma](c, timeperiod = int( optimized_period[ma])))
low_vol = low_vol.fillna(0)
high_vol = pd.DataFrame()
df2 = df.copy()
for i in df2.columns:
if i in low_vol.columns:
high_vol[i] = df2[i].subtract(low_vol[i], fill_value=0)
SMA
SMA([input_arrays], [timeperiod=30])
Simple Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
17
EMA
EMA([input_arrays], [timeperiod=30])
Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
51
WMA
WMA([input_arrays], [timeperiod=30])
Weighted Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
49
DEMA
DEMA([input_arrays], [timeperiod=30])
Double Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
89
KAMA
KAMA([input_arrays], [timeperiod=30])
Kaufman Adaptive Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
18
MIDPOINT
MIDPOINT([input_arrays], [timeperiod=14])
MidPoint over period (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 14
Outputs:
real
14
T3
T3([input_arrays], [timeperiod=5], [vfactor=0.7])
Triple Exponential Moving Average (T3) (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 5
vfactor: 0.7
Outputs:
real
19
TEMA
TEMA([input_arrays], [timeperiod=30])
Triple Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
9
In [ ]:
low_vol.tail(20)
Out[ ]:
| open | high | low | close | Adj Close | volume | MA7 | MA21 | MACD | 20SD | upper_band | lower_band | EMA | logmomentum | absolute of 3 comp | angle of 3 comp | absolute of 6 comp | angle of 6 comp | absolute of 9 comp | angle of 9 comp | search | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1060 | 140.200839 | 141.942909 | 138.524500 | 140.171495 | 139.966842 | 8.852448e+07 | 142.165478 | 146.699207 | 1.815578 | 4.572948 | 155.845103 | 137.553312 | 140.365562 | 4.935800 | 105.739092 | -0.047411 | 125.318767 | -0.018291 | 140.471430 | -0.008749 | 19.573385 |
| 1061 | 139.425914 | 141.705469 | 138.035200 | 140.698014 | 140.492650 | 8.620711e+07 | 141.528981 | 145.978836 | 2.115887 | 4.189393 | 154.357621 | 137.600050 | 140.587196 | 4.939545 | 105.263514 | -0.048037 | 124.464999 | -0.019222 | 139.335869 | -0.009472 | 19.632022 |
| 1062 | 140.773058 | 142.636405 | 139.932338 | 141.733666 | 141.526843 | 7.421445e+07 | 141.294887 | 145.298477 | 2.211018 | 3.647690 | 152.593858 | 138.003097 | 141.351509 | 4.946870 | 104.786174 | -0.048658 | 123.598217 | -0.020150 | 138.164839 | -0.010188 | 19.698090 |
| 1063 | 142.179695 | 143.266994 | 141.127848 | 142.249061 | 142.041527 | 6.519616e+07 | 141.224295 | 144.665584 | 2.093072 | 3.241276 | 151.148137 | 138.183031 | 141.949877 | 4.950518 | 104.307114 | -0.049275 | 122.718682 | -0.021074 | 136.959041 | -0.010898 | 19.763505 |
| 1064 | 142.253947 | 144.008334 | 141.546689 | 142.555532 | 142.347589 | 6.254214e+07 | 141.336839 | 144.184381 | 1.988881 | 2.884864 | 149.954110 | 138.414652 | 142.353647 | 4.952685 | 103.826381 | -0.049886 | 121.826667 | -0.021994 | 135.719217 | -0.011600 | 20.311676 |
| 1065 | 142.782738 | 143.732491 | 141.438660 | 142.125353 | 141.918068 | 6.542511e+07 | 141.385297 | 143.758659 | 1.774804 | 2.626682 | 149.012024 | 138.505294 | 142.201451 | 4.949632 | 103.344020 | -0.050491 | 120.922446 | -0.022909 | 134.446150 | -0.012293 | 20.671514 |
| 1066 | 142.153085 | 142.656915 | 140.466684 | 141.564232 | 141.357788 | 7.040262e+07 | 141.585336 | 143.387397 | 1.634667 | 2.376817 | 148.141030 | 138.633764 | 141.776638 | 4.945637 | 102.860075 | -0.051092 | 120.006305 | -0.023818 | 133.140665 | -0.012977 | 20.900131 |
| 1067 | 142.177201 | 143.194327 | 140.977156 | 142.610382 | 142.402435 | 6.948112e+07 | 141.933749 | 143.094536 | 1.573317 | 2.074153 | 147.242842 | 138.946230 | 142.332468 | 4.953023 | 102.374593 | -0.051687 | 119.078535 | -0.024722 | 131.803627 | -0.013650 | 21.038585 |
| 1068 | 143.009006 | 144.052615 | 142.286776 | 143.812497 | 143.602819 | 6.805244e+07 | 142.378675 | 142.879716 | 1.473333 | 1.874158 | 146.628032 | 139.131400 | 143.319154 | 4.961467 | 101.887619 | -0.052275 | 118.139433 | -0.025618 | 130.435938 | -0.014311 | 21.116168 |
| 1069 | 143.380322 | 145.547752 | 142.940349 | 145.397429 | 145.185452 | 7.592729e+07 | 142.902069 | 142.813890 | 1.447641 | 1.844159 | 146.502207 | 139.125573 | 144.704671 | 4.972505 | 101.399198 | -0.052858 | 117.189304 | -0.026508 | 129.038540 | -0.014959 | 21.153587 |
| 1070 | 145.337970 | 147.615882 | 144.980528 | 147.444584 | 147.229635 | 7.653090e+07 | 143.644287 | 142.961273 | 1.284466 | 2.010227 | 146.981728 | 138.940819 | 146.531280 | 4.986604 | 100.909377 | -0.053435 | 116.228458 | -0.027389 | 127.612408 | -0.015592 | 21.165321 |
| 1071 | 147.375283 | 149.163050 | 146.995423 | 148.921380 | 148.704294 | 6.811986e+07 | 144.553694 | 143.236380 | 0.961952 | 2.270386 | 147.777152 | 138.695607 | 148.124680 | 4.996737 | 100.418203 | -0.054006 | 115.257214 | -0.028261 | 126.158555 | -0.016211 | 21.161363 |
| 1072 | 148.656821 | 150.010875 | 148.071943 | 149.870634 | 149.652170 | 6.425222e+07 | 145.660163 | 143.530869 | 0.589081 | 2.556352 | 148.643574 | 138.418164 | 149.288649 | 5.003230 | 99.925720 | -0.054570 | 114.275894 | -0.029124 | 124.678027 | -0.016812 | 21.148490 |
| 1073 | 149.806550 | 150.715254 | 149.026204 | 149.977942 | 149.759331 | 6.069918e+07 | 146.862121 | 143.785380 | 0.135134 | 2.805932 | 149.397244 | 138.173516 | 149.748178 | 5.003989 | 99.431976 | -0.055128 | 113.284828 | -0.029977 | 123.171903 | -0.017396 | 21.131204 |
| 1074 | 149.937482 | 150.666013 | 149.022091 | 149.911667 | 149.693162 | 5.465321e+07 | 147.905162 | 144.001463 | -0.245163 | 3.045742 | 150.092948 | 137.909978 | 149.857170 | 5.003545 | 98.937018 | -0.055679 | 112.284350 | -0.030820 | 121.641290 | -0.017961 | 25.016406 |
| 1075 | 150.228161 | 151.254072 | 149.586503 | 150.104281 | 149.885502 | 5.602702e+07 | 148.803988 | 144.237215 | -0.571069 | 3.270011 | 150.777237 | 137.697192 | 150.021910 | 5.004835 | 98.440892 | -0.056223 | 111.274800 | -0.031650 | 120.087330 | -0.018506 | 27.455491 |
| 1076 | 150.328251 | 150.997797 | 149.591175 | 149.912656 | 149.694163 | 5.484778e+07 | 149.449021 | 144.548659 | -0.850904 | 3.458615 | 151.465890 | 137.631428 | 149.949074 | 5.003520 | 97.943645 | -0.056759 | 110.256524 | -0.032469 | 118.511190 | -0.019029 | 28.912854 |
| 1077 | 150.525566 | 152.430694 | 150.099878 | 151.531571 | 151.310718 | 7.580033e+07 | 150.032876 | 144.967153 | -0.975625 | 3.719924 | 152.407001 | 137.527305 | 151.004072 | 5.014296 | 97.445324 | -0.057289 | 109.229873 | -0.033274 | 116.914063 | -0.019528 | 29.716707 |
| 1078 | 149.301052 | 151.688142 | 148.723104 | 151.137179 | 150.916905 | 1.012990e+08 | 150.349418 | 145.413317 | -0.891585 | 3.905336 | 153.223988 | 137.602646 | 151.092810 | 5.011652 | 96.945977 | -0.057811 | 108.195203 | -0.034066 | 115.297171 | -0.020004 | 30.096629 |
| 1079 | 149.321425 | 151.018197 | 148.455004 | 150.396057 | 150.176865 | 9.262134e+07 | 150.424479 | 145.823313 | -0.852689 | 3.878291 | 153.579894 | 138.066731 | 150.628308 | 5.006660 | 96.445650 | -0.058325 | 107.152874 | -0.034844 | 113.661756 | -0.020453 | 27.283213 |
In [ ]:
high_vol.head(10)
Out[ ]:
| open | high | low | close | Adj Close | volume | MA7 | MA21 | MACD | 20SD | upper_band | lower_band | EMA | logmomentum | absolute of 3 comp | angle of 3 comp | absolute of 6 comp | angle of 6 comp | absolute of 9 comp | angle of 9 comp | search | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 36.220001 | 36.325001 | 35.775002 | 35.875000 | 34.054882 | 57111200.0 | 36.173571 | 36.751904 | 0.303356 | 0.960520 | 38.672945 | 34.830864 | 35.924548 | 3.551770 | 38.458011 | 0.046984 | 29.704545 | 0.102857 | 43.304973 | -0.053955 | 15.0 |
| 1 | 35.922501 | 36.197498 | 35.680000 | 36.022499 | 34.194897 | 86278400.0 | 36.095357 | 36.634762 | 0.328795 | 0.852735 | 38.340231 | 34.929292 | 35.989849 | 3.555991 | 38.240991 | 0.049445 | 29.954520 | 0.099254 | 43.438321 | -0.053936 | 15.0 |
| 2 | 35.755001 | 35.875000 | 35.602501 | 35.682499 | 33.872143 | 96515200.0 | 35.984999 | 36.495238 | 0.346702 | 0.677629 | 37.850495 | 35.139980 | 35.784949 | 3.546235 | 38.027974 | 0.051918 | 30.209839 | 0.095602 | 43.557403 | -0.053820 | 15.0 |
| 3 | 35.724998 | 36.187500 | 35.724998 | 36.044998 | 34.216255 | 76806800.0 | 36.001071 | 36.362023 | 0.387422 | 0.387634 | 37.137291 | 35.586756 | 35.958315 | 3.556633 | 37.818962 | 0.054401 | 30.470232 | 0.091907 | 43.662260 | -0.053608 | 15.0 |
| 4 | 36.027500 | 36.487499 | 35.842499 | 36.264999 | 34.425095 | 84362400.0 | 35.973571 | 36.243809 | 0.388315 | 0.308042 | 36.859893 | 35.627725 | 36.162771 | 3.562891 | 37.613953 | 0.056893 | 30.735430 | 0.088177 | 43.752965 | -0.053302 | 14.0 |
| 5 | 36.182499 | 36.462502 | 36.095001 | 36.382500 | 34.536625 | 79127200.0 | 36.039642 | 36.202738 | 0.372153 | 0.308860 | 36.820458 | 35.585018 | 36.309257 | 3.566217 | 37.412947 | 0.059392 | 31.005161 | 0.084416 | 43.829622 | -0.052901 | 14.0 |
| 6 | 36.467499 | 36.544998 | 36.205002 | 36.435001 | 34.586472 | 99538000.0 | 36.101071 | 36.206547 | 0.317572 | 0.295861 | 36.798268 | 35.614826 | 36.393086 | 3.567700 | 37.215939 | 0.061899 | 31.279154 | 0.080632 | 43.892360 | -0.052406 | 14.0 |
| 7 | 36.375000 | 37.122501 | 36.360001 | 36.942501 | 35.068211 | 100797600.0 | 36.253571 | 36.220595 | 0.322643 | 0.340687 | 36.901969 | 35.539221 | 36.759363 | 3.581920 | 37.022928 | 0.064410 | 31.557136 | 0.076830 | 43.941338 | -0.051818 | 14.0 |
| 8 | 36.992500 | 37.332500 | 36.832500 | 37.259998 | 35.369610 | 80528400.0 | 36.430357 | 36.266785 | 0.257925 | 0.410484 | 37.087753 | 35.445818 | 37.093120 | 3.590715 | 36.833908 | 0.066926 | 31.838833 | 0.073014 | 43.976744 | -0.051137 | 14.0 |
| 9 | 37.205002 | 37.724998 | 37.142502 | 37.389999 | 35.493000 | 95174000.0 | 36.674285 | 36.329523 | 0.184267 | 0.445597 | 37.220717 | 35.438330 | 37.291039 | 3.594294 | 36.648875 | 0.069445 | 32.123972 | 0.069192 | 43.998789 | -0.050365 | 16.0 |
Common Functions¶
In [77]:
def get_arima(dataframe,original_data, train_len, test_len):
# prepare train and test data
X_value = pd.DataFrame(dataframe.iloc[:, :])
y_value = pd.DataFrame(dataframe.iloc[:, 3])
X_train, X_test = split_train_test(X_value)
y_train, y_test = split_train_test(y_value)
yc_train,yc_test = split_train_test(original_data)
# y_train_ = y_train['close'].to_list()
# y_test_ = y_test['close'].to_list()
yc = yc_test.values.tolist()
y_train_list = y_train['close'].values.tolist()
y_test_list = y_test['close'].values.tolist()
# Initialize model
model = auto_arima(y_train_list,trace=True, error_action='ignore', start_p=1,start_q=1,max_p=3,max_q=3,d=3,
suppress_warnings=True,stepwise=True,seasonal=True)
print(model.summary())
# Determine model parameters
model.fit(y_train_list,disp= 0)
order = model.get_params()['order']
print('ARIMA order:', order, '\n')
# Genereate predictions
prediction = []
for i in range(len(y_test_list)):
model = pmdarima.ARIMA(order=order)
model.fit(y_train_list,disp= 0)
# print('working on', i+1, 'of', len(y_test), '-- ' + str(int(100 * (i + 1) / len(y_test))) + '% complete')
prediction.append(model.predict()[0])
y_train_list.append(y_test_list[i])
# Generate error data
mse = mean_squared_error(yc_test, prediction)
rmse = mse ** 0.5
# mape = mean_absolute_percentage_error(pd.Series(yc_test).values.tolist(), pd.Series(predictionte).values.tolist() )
mae = mean_absolute_error(pd.Series(yc_test).values.tolist() , pd.Series(prediction).values.tolist() )
return yc, prediction, mse, rmse, mae
In [95]:
def plot_train(simulation,SIM):
train_predict_index = np.load("index_train_appl.npy", allow_pickle=True)#Dates for train data
predict_result = pd.DataFrame()
for i in range(len(simulation[SIM]['final_tr']['prediction'])):
y_predict = pd.DataFrame(simulation[SIM]['final_tr']['prediction'][i], columns=["predicted_price"],
index=train_predict_index[i:i + output_dim])
predict_result = pd.concat([predict_result, y_predict], axis=1, sort=False)
#This is a dataframe with each column containing the predicted daily closing price
real_price = pd.DataFrame()
for i in range(len(simulation[SIM]['final_tr']['original'])):
y_train = pd.DataFrame(simulation[SIM]['final_tr']['original'][i], columns=["real_price"],
index=train_predict_index[i:i + output_dim])
real_price = pd.concat([real_price, y_train], axis=1, sort=False) #This is a dataframe with each column containing the real daily closing price
predict_result['predicted_mean'] = predict_result.mean(axis=1)#Adding a column with the daily predicted closing price value
real_price['real_mean'] = real_price.mean(axis=1)#Adding a column with the daily real closing price value
#
# Plot the predicted result
plt.figure(figsize=(16, 8))
plt.plot(real_price["real_mean"])
plt.plot(predict_result["predicted_mean"], color='r')
plt.xlabel("Date")
plt.ylabel("Stock price")
plt.legend(("Real price", "Predicted price"), loc="upper left", fontsize=16)
plt.title(f"The result of Training for Hybrid Arima LSTM with MA - {SIM} : {fileimg}",fontsize=20)
plt.show()
# Calculate RMSE
predicted = predict_result["predicted_mean"]
real = real_price["real_mean"]
RMSE = np.sqrt(mean_squared_error(predicted, real))
MSE = mean_squared_error(predicted, real)
MAE = mean_absolute_error(predicted, real)
print(f"----- Train RMSE for {SIM} -----", RMSE)
print(f"----- Train_MSE_LSTM for {SIM} -----", MSE)
print(f"----- Train MAE LSTM for {SIM} -----", MAE)
In [96]:
def plot_test(simulation, SIM):
test_predict_index = np.load("index_test_appl.npy", allow_pickle=True)#Dates for train data
# rescaled_real_y = y_scaler.inverse_transform(y_train)#Real closing price data
# rescaled_predicted_y = y_scaler.inverse_transform(train_yhat)#Predicted closing price data
predict_result = pd.DataFrame()
for i in range(len(simulation[SIM]['final']['prediction'])):
y_predict = pd.DataFrame(simulation[SIM]['final']['prediction'][i], columns=["predicted_price"],
index=test_predict_index[i:i + output_dim])
predict_result = pd.concat([predict_result, y_predict], axis=1, sort=False)#This is a dataframe with each column containing the predicted daily closing price
#
real_price = pd.DataFrame()
for i in range(len(simulation[SIM]['final']['original'])):
y_train = pd.DataFrame(simulation[SIM]['final']['original'][i], columns=["real_price"],
index=test_predict_index[i:i + output_dim])
real_price = pd.concat([real_price, y_train], axis=1, sort=False)#This is a dataframe with each column containing the real daily closing price
predict_result['predicted_mean'] = predict_result.mean(axis=1)#Adding a column with the daily predicted closing price value
real_price['real_mean'] = real_price.mean(axis=1)#Adding a column with the daily real closing price value
#
# Plot the predicted result
plt.figure(figsize=(16, 8))
plt.plot(real_price["real_mean"])
plt.plot(predict_result["predicted_mean"], color='r')
plt.xlabel("Date")
plt.ylabel("Stock price")
plt.legend(("Real price", "Predicted price"), loc="upper left", fontsize=16)
plt.title(f"The result of Testing for Hybrid Arima LSTM with MA - {SIM} : {fileimg}",fontsize=20)
plt.show()
# Calculate RMSE
predicted = predict_result["predicted_mean"]
real = real_price["real_mean"]
RMSE = np.sqrt(mean_squared_error(predicted, real))
MSE = mean_squared_error(predicted, real)
MAE = mean_absolute_error(predicted, real)
print(f"----- Test RMSE for {SIM}-----", RMSE)
print(f"----- Test_MSE_LSTM for {SIM}-----", MSE)
print(f"----- Test_MAE_LSTM for {SIM}-----", MAE)
In [10]:
def plot_train_high(simulation, SIM):
train_predict_index = np.load("index_test_appl.npy", allow_pickle=True)#Dates for train data
predict_result = pd.DataFrame()
for i in range(len(simulation[SIM]['high_vol']['prediction'])):
y_predict = pd.DataFrame(simulation[SIM]['high_vol']['prediction'][i], columns=["predicted_price"],
index=train_predict_index[i:i + output_dim])
predict_result = pd.concat([predict_result, y_predict], axis=1, sort=False)
#This is a dataframe with each column containing the predicted daily closing price
real_price = pd.DataFrame()
for i in range(len(simulation[SIM]['high_vol']['original'])):
y_train = pd.DataFrame(simulation[SIM]['high_vol']['original'][i], columns=["real_price"],
index=train_predict_index[i:i + output_dim])
real_price = pd.concat([real_price, y_train], axis=1, sort=False) #This is a dataframe with each column containing the real daily closing price
predict_result['predicted_mean'] = predict_result.mean(axis=1)#Adding a column with the daily predicted closing price value
real_price['real_mean'] = real_price.mean(axis=1)#Adding a column with the daily real closing price value
#
# Plot the predicted result
plt.figure(figsize=(16, 8))
plt.plot(real_price["real_mean"])
plt.plot(predict_result["predicted_mean"], color='r')
plt.xlabel("Date")
plt.ylabel("Stock price")
plt.legend(("Real price", "Predicted price"), loc="upper left", fontsize=16)
plt.title(f"The result of Training for Hybrid Arima LSTM with MA {SIM}", fontsize=20)
plt.show()
# Calculate RMSE
predicted = predict_result["predicted_mean"]
real = real_price["real_mean"]
RMSE = np.sqrt(mean_squared_error(predicted, real))
MSE = mean_squared_error(predicted, real)
MAE = mean_absolute_error(predicted, real)
print(f"----- Individual LSTM RMSE for {SIM} -----", RMSE)
print(f"----- Individual LSTM_MSE_LSTM for {SIM} -----", MSE)
print(f"----- Individual LSTM MAE LSTM for {SIM} -----", MAE)
In [81]:
def plot_train_low(simulation , SIM):
train_predict_index = np.load("index_test_appl.npy", allow_pickle=True)#Dates for train data
predict_result = pd.DataFrame()
for i in range(len(simulation[SIM]['low_vol']['prediction'])):
y_predict = pd.DataFrame(simulation[SIM]['low_vol']['prediction'][i], columns=["predicted_price"],
index=train_predict_index[i:i + output_dim])
predict_result = pd.concat([predict_result, y_predict], axis=1, sort=False)
#This is a dataframe with each column containing the predicted daily closing price
real_price = pd.DataFrame()
for i in range(len(simulation[SIM]['low_vol']['original'])):
y_train = pd.DataFrame(simulation[SIM]['low_vol']['original'][i], columns=["real_price"],
index=train_predict_index[i:i + output_dim])
real_price = pd.concat([real_price, y_train], axis=1, sort=False) #This is a dataframe with each column containing the real daily closing price
predict_result['predicted_mean'] = predict_result.mean(axis=1)#Adding a column with the daily predicted closing price value
real_price['real_mean'] = real_price.mean(axis=1)#Adding a column with the daily real closing price value
#
# Plot the predicted result
plt.figure(figsize=(16, 8))
plt.plot(real_price["real_mean"])
plt.plot(predict_result["predicted_mean"], color='r')
plt.xlabel("Date")
plt.ylabel("Stock price")
plt.legend(("Real price", "Predicted price"), loc="upper left", fontsize=16)
plt.title(f"The result of Training for {SIM}", fontsize=20)
plt.show()
# Calculate RMSE
predicted = predict_result["predicted_mean"]
real = real_price["real_mean"]
RMSE = np.sqrt(mean_squared_error(predicted, real))
MSE = mean_squared_error(predicted, real)
MAE = mean_absolute_error(predicted, real)
print(f"-----Arima RMSE for {SIM} -----", RMSE)
print(f"----- Arima MSE for {SIM} -----", MSE)
print(f"----- Arima MAE for {SIM} -----", MAE)
In [82]:
import os
os.getcwd()
Out[82]:
'/content/drive/.shortcut-targets-by-id/1IaGjVBlTspxI2CHSrxfYnaiYvsaG0pHs/Stock price prediction/Archana - LSTM Hybrid/Outputs/sentiment'
Univariate Arima Multistep MutiVariate LSTM Hybrid Model Experiment 1¶
In [83]:
def get_lstm(data,original_data, train_len, test_len,img_file,ma ,lstm_len=3):
# prepare train and test data
X_value = pd.DataFrame(data.iloc[:, :])
y_value = pd.DataFrame(data.iloc[:, 3])
X_scaler = MinMaxScaler(feature_range=(-1, 1))
y_scaler = MinMaxScaler(feature_range=(-1, 1))
X_scaler.fit(X_value)
y_scaler.fit(y_value)
# Get data and check shape
X, y, yc = get_X_y(X_scale_dataset, y_scale_dataset)#X will be of shape 224 X 3 X 21 (each 3 X 21 array will be 3 days' worth of data). yc will have the corresponding closing price value
# pdb.set_trace()
X_train, X_test, = split_train_test(X)
y_train, y_test, = split_train_test(y)
# yc_train, yc_test, = split_train_test(original_data)
index_train, index_test, = predict_index(dataset_final, X_train, n_steps_in, n_steps_out)
det = 20
input_dim = X_train.shape[1]#3
feature_size = X_train.shape[2]#24
output_dim = y_train.shape[1]#1
# Option 1
# Set up & fit LSTM RNN
model = Sequential()
model.add(LSTM(256, activation='relu', kernel_initializer='he_normal', input_shape=(input_dim, feature_size)))
model.add(Dense(units=64,activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(units=output_dim))
model.compile(optimizer=Adam(learning_rate = 0.001), loss='mse')
## Common code
callbacks = [
EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
fname1 = img_file+'.png'
tensorflow.keras.utils.plot_model(
model, to_file=fname1, show_shapes=True, show_dtype=False,
show_layer_names=True, expand_nested=False, dpi=96,
layer_range=None, show_layer_activations=False
)
history = model.fit(X_train, y_train, epochs=500, batch_size=int( optimized_period[ma]), verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# plot loss
fname2 = img_file+'-'+ma
plt.title(img_file+'-'+ma+' Loss')
plt.xlabel("Epochs")
plt.ylabel("Loss")
pyplot.plot(history.history['loss'], label='train')
pyplot.plot(history.history['val_loss'], label='validation')
pyplot.legend()
pyplot.savefig(fname2+'.png',dpi='figure')
pyplot.show()
# # option 2
# model = Sequential()
# model.add(Bidirectional(LSTM(units= 128), input_shape=(input_dim, feature_size)))
# model.add(Dense(64))
# model.add(Dense(units=output_dim))
# model.compile(optimizer=Adam(lr = 0.001), loss='mean_squared_error', metrics=['accuracy'])
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# Option 3
# define custom activation
# cts().update({'double_tanh':Double_Tanh(double_tanh)})
# # Model Generation
# model = Sequential()
# #check https://machinelearningmastery.com/use-weight-regularization-lstm-networks-time-series-forecasting/
# model.add(LSTM(25, input_shape=(input_dim, feature_size), dropout=0.2, kernel_regularizer=l1_l2(0.00,0.00), bias_regularizer=l1_l2(0.00,0.00)))
# model.add(Dense(1))
# model.add(Activation(double_tanh))
# model.compile(loss='mean_squared_error', optimizer='adam', metrics=['mse', 'mae'])
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# Option 4
# Set up & fit LSTM RNN
# model = Sequential()
# model.add(LSTM(units=lstm_len, return_sequences=True, input_shape=(x_train.shape[1], 1)))
# model.add(LSTM(units=int(lstm_len/2)))
# model.add(Dense(1, activation='sigmoid'))
# model.compile(loss='mean_squared_error', optimizer='adam')
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# Generate predictions
predictiontr = model.predict(X_train, verbose=0)
predictiontr = y_scaler.inverse_transform(predictiontr).tolist()
outputtr = []
for i in range(len(predictiontr)):
outputtr.extend(predictiontr[i])
predictiontr = outputtr
# Generate error data
## replace with yc , xtest generated by new multistep method
mse_tr = mean_squared_error(y_train, predictiontr)
rmse_tr = mse_tr ** 0.5
# mape = mean_absolute_percentage_error(X_test, pd.Series(predictiontr))
mae_tr = mean_absolute_error(y_train, pd.Series(predictiontr))
# Original_tr = pd.Series(yc_train)
Original_tr = y_scaler.inverse_transform(y_train).flatten().tolist()
predictionte = model.predict(X_test, verbose=0)
predictionte = (y_scaler.inverse_transform(predictionte)-det).tolist()
outputte = []
for i in range(len(predictionte)):
outputte.extend(predictionte[i])
predictionte = outputte
# Generate error data
mse_te = mean_squared_error(y_test, predictionte)
rmse_te = mse_te ** 0.5
# mape = mean_absolute_percentage_error(X_test, pd.Series(predictionte))
mae_te = mean_absolute_error(y_test, pd.Series(predictionte))
# Original_te = pd.Series(yc_test)
Original_te = y_scaler.inverse_transform(y_test).flatten().tolist()
return Original_tr, predictiontr, mse_tr, rmse_tr,mae_tr,Original_te,predictionte, mse_te,rmse_te,mae_te
In [84]:
if __name__ == '__main__':
start_time = timeit.default_timer()
simulation1 = {}
imgfile = 'Experiment1'
for ma in optimized_period:
print(ma)
print(functions[ma])
print ( int( optimized_period[ma]))
# if ma == 'SMA':
low_vol = df.apply(lambda c: functions[ma](c, timeperiod = int( optimized_period[ma])))
low_vol = low_vol.fillna(0)
low_vol_data = df['close']
high_vol = pd.DataFrame()
df2 = df.copy()
for i in df2.columns:
if i in low_vol.columns:
high_vol[i] = df2[i].subtract(low_vol[i], fill_value=0)
high_vol_data = df['close']
## *****************************************************
# Generate ARIMA and LSTM predictions
print('\nWorking on ' + ma + ' predictions')
try:
print('parameters used : ', train_len, test_len)
low_vol_Original, low_vol_prediction, low_vol_mse, low_vol_rmse,low_vol_mae = get_arima(low_vol,low_vol_data, train_len, test_len)
except:
print('ARIMA error, skipping to next MA type')
continue
Original_tr, predictiontr, mse_tr, rmse_tr,mae_tr,high_vol_Original, high_vol_prediction, high_vol_mse, high_vol_rmse,high_vol_mae, = get_lstm(high_vol,high_vol_data, train_len, test_len,imgfile,ma)
final_prediction_tr = df['close'].head(train_len).values + pd.Series(predictiontr) # ignoring first 3 steps
mse_ftr = mean_squared_error(df['close'].head(train_len).values,final_prediction_tr.values)
rmse_ftr = mse_ftr ** 0.5
mape_ftr = mean_absolute_percentage_error(df['close'].head(train_len).reset_index(drop=True), final_prediction_tr)
mae_ftr = mean_absolute_error(df['close'].head(train_len).reset_index(drop=True), final_prediction_tr)
final_prediction = pd.Series(low_vol_prediction[3:]) + pd.Series(high_vol_prediction)
mse = mean_squared_error(df['close'].tail(test_len).values,final_prediction.values)
rmse = mse ** 0.5
mape = mean_absolute_percentage_error(df['close'].tail(test_len).reset_index(drop=True), final_prediction)
mae = mean_absolute_error(df['close'].tail(test_len).reset_index(drop=True), final_prediction)
# Generate prediction accuracy
actual = df['close'].tail(test_len).values
result_1 = []
result_2 = []
for i in range(1, len(final_prediction)):
# Compare prediction to previous close price
if final_prediction[i] > actual[i-1] and actual[i] > actual[i-1]:
result_1.append(1)
elif final_prediction[i] < actual[i-1] and actual[i] < actual[i-1]:
result_1.append(1)
else:
result_1.append(0)
# Compare prediction to previous prediction
if final_prediction[i] > final_prediction[i-1] and actual[i] > actual[i-1]:
result_2.append(1)
elif final_prediction[i] < final_prediction[i-1] and actual[i] < actual[i-1]:
result_2.append(1)
else:
result_2.append(0)
accuracy_1 = np.mean(result_1)
accuracy_2 = np.mean(result_2)
simulation1[ma] = {'low_vol': {'original':list(low_vol_Original), 'prediction': list(low_vol_prediction) , 'mse': low_vol_mse,
'rmse': low_vol_rmse, 'mae' : low_vol_mae},
'high_vol': {'original':list(high_vol_Original),'prediction': list(high_vol_prediction), 'mse': high_vol_mse,
'rmse': high_vol_rmse, 'mae' : high_vol_mae},
'final_tr': {'original':df['close'].head(train_len).tolist(),'prediction': final_prediction_tr.values.tolist(), 'mse': mse_ftr,
'rmse': rmse_ftr, 'mae' : mae_ftr},
'final': {'original': df['close'].tail(test_len).tolist(), 'prediction': final_prediction.values.tolist(), 'mse': mse,
'rmse': rmse, 'mae': mae },
'accuracy': {'prediction vs close': accuracy_1, 'prediction vs prediction': accuracy_2}}
# save simulation data here as checkpoint
with open('simulation1_data.json', 'w') as fp:
json.dump(simulation1, fp)
for ma in simulation1.keys():
print('\n' + ma)
print('Prediction vs Close:\t\t' + str(round(100*simulation1[ma]['accuracy']['prediction vs close'], 2))
+ '% Accuracy')
print('Prediction vs Prediction:\t' + str(round(100*simulation1[ma]['accuracy']['prediction vs prediction'], 2))
+ '% Accuracy')
print('MSE:\t', simulation1[ma]['final']['mse'],
'\nRMSE:\t', simulation1[ma]['final']['rmse'],
'\nMAPE:\t', simulation1[ma]['final']['mae'])#,
# '\nMAPE:\t', simulation[ma]['final']['mape'])
# else:
# break
# code you want to evaluate
elapsed = timeit.default_timer() - start_time
print('Runtime: mins:',elapsed/60)
SMA
SMA([input_arrays], [timeperiod=30])
Simple Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
17
Working on SMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.54 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4157.020, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3687.148, Time=0.05 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.20 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3458.651, Time=0.08 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3322.133, Time=0.11 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=0.80 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.88 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3324.133, Time=0.22 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 2.922 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1657.067
Date: Sun, 12 Dec 2021 AIC 3322.133
Time: 13:22:03 BIC 3340.897
Sample: 0 HQIC 3329.339
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1966 0.003 -387.226 0.000 -1.203 -1.191
ar.L2 -0.8952 0.006 -138.692 0.000 -0.908 -0.883
ar.L3 -0.3968 0.006 -68.284 0.000 -0.408 -0.385
sigma2 3.5858 0.017 214.535 0.000 3.553 3.619
===================================================================================
Ljung-Box (L1) (Q): 14.47 Jarque-Bera (JB): 2428881.42
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.90
Prob(H) (two-sided): 0.00 Kurtosis: 271.99
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
WARNING:tensorflow:Layer lstm will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
WARNING:tensorflow:Layer lstm will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.16778, saving model to LSTM1.h5 48/48 - 3s - loss: 0.1712 - val_loss: 0.1678 - lr: 0.0010 - 3s/epoch - 62ms/step Epoch 2/500 Epoch 00002: val_loss did not improve from 0.16778 48/48 - 0s - loss: 0.1883 - val_loss: 0.3040 - lr: 0.0010 - 419ms/epoch - 9ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.16778 48/48 - 1s - loss: 0.0672 - val_loss: 0.5210 - lr: 0.0010 - 519ms/epoch - 11ms/step Epoch 4/500 Epoch 00004: val_loss did not improve from 0.16778 48/48 - 0s - loss: 0.0577 - val_loss: 0.2578 - lr: 0.0010 - 407ms/epoch - 8ms/step Epoch 5/500 Epoch 00005: val_loss improved from 0.16778 to 0.16001, saving model to LSTM1.h5 48/48 - 0s - loss: 0.0489 - val_loss: 0.1600 - lr: 0.0010 - 456ms/epoch - 9ms/step Epoch 6/500 Epoch 00006: val_loss improved from 0.16001 to 0.07747, saving model to LSTM1.h5 48/48 - 0s - loss: 0.0417 - val_loss: 0.0775 - lr: 0.0010 - 478ms/epoch - 10ms/step Epoch 7/500 Epoch 00007: val_loss improved from 0.07747 to 0.03761, saving model to LSTM1.h5 48/48 - 0s - loss: 0.0437 - val_loss: 0.0376 - lr: 0.0010 - 452ms/epoch - 9ms/step Epoch 8/500 Epoch 00008: val_loss improved from 0.03761 to 0.01788, saving model to LSTM1.h5 48/48 - 0s - loss: 0.0414 - val_loss: 0.0179 - lr: 0.0010 - 445ms/epoch - 9ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.01788 48/48 - 0s - loss: 0.0381 - val_loss: 0.0802 - lr: 0.0010 - 469ms/epoch - 10ms/step Epoch 10/500 Epoch 00010: val_loss improved from 0.01788 to 0.00515, saving model to LSTM1.h5 48/48 - 0s - loss: 0.0452 - val_loss: 0.0051 - lr: 0.0010 - 380ms/epoch - 8ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.00515 48/48 - 0s - loss: 0.0571 - val_loss: 0.0147 - lr: 0.0010 - 430ms/epoch - 9ms/step Epoch 12/500 Epoch 00012: val_loss improved from 0.00515 to 0.00495, saving model to LSTM1.h5 48/48 - 0s - loss: 0.0581 - val_loss: 0.0050 - lr: 0.0010 - 433ms/epoch - 9ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0547 - val_loss: 0.0245 - lr: 0.0010 - 368ms/epoch - 8ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0348 - val_loss: 0.0158 - lr: 0.0010 - 404ms/epoch - 8ms/step Epoch 15/500 Epoch 00015: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0303 - val_loss: 0.3462 - lr: 0.0010 - 438ms/epoch - 9ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.00495 48/48 - 1s - loss: 0.0273 - val_loss: 0.1067 - lr: 0.0010 - 504ms/epoch - 11ms/step Epoch 17/500 Epoch 00017: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00017: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0265 - val_loss: 0.0396 - lr: 0.0010 - 439ms/epoch - 9ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0289 - val_loss: 0.0407 - lr: 1.0000e-04 - 402ms/epoch - 8ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0249 - val_loss: 0.0413 - lr: 1.0000e-04 - 405ms/epoch - 8ms/step Epoch 20/500 Epoch 00020: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0251 - val_loss: 0.0442 - lr: 1.0000e-04 - 396ms/epoch - 8ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0251 - val_loss: 0.0440 - lr: 1.0000e-04 - 434ms/epoch - 9ms/step Epoch 22/500 Epoch 00022: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00022: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0253 - val_loss: 0.0419 - lr: 1.0000e-04 - 358ms/epoch - 7ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0254 - val_loss: 0.0413 - lr: 1.0000e-05 - 408ms/epoch - 9ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0232 - val_loss: 0.0415 - lr: 1.0000e-05 - 492ms/epoch - 10ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0216 - val_loss: 0.0410 - lr: 1.0000e-05 - 444ms/epoch - 9ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0225 - val_loss: 0.0413 - lr: 1.0000e-05 - 442ms/epoch - 9ms/step Epoch 27/500 Epoch 00027: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00027: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0239 - val_loss: 0.0411 - lr: 1.0000e-05 - 470ms/epoch - 10ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0229 - val_loss: 0.0413 - lr: 1.0000e-05 - 449ms/epoch - 9ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0239 - val_loss: 0.0417 - lr: 1.0000e-05 - 386ms/epoch - 8ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0218 - val_loss: 0.0418 - lr: 1.0000e-05 - 448ms/epoch - 9ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0213 - val_loss: 0.0420 - lr: 1.0000e-05 - 429ms/epoch - 9ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0228 - val_loss: 0.0420 - lr: 1.0000e-05 - 482ms/epoch - 10ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0234 - val_loss: 0.0415 - lr: 1.0000e-05 - 463ms/epoch - 10ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0225 - val_loss: 0.0423 - lr: 1.0000e-05 - 423ms/epoch - 9ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0239 - val_loss: 0.0423 - lr: 1.0000e-05 - 437ms/epoch - 9ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0226 - val_loss: 0.0420 - lr: 1.0000e-05 - 441ms/epoch - 9ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0222 - val_loss: 0.0413 - lr: 1.0000e-05 - 410ms/epoch - 9ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0233 - val_loss: 0.0408 - lr: 1.0000e-05 - 421ms/epoch - 9ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0231 - val_loss: 0.0411 - lr: 1.0000e-05 - 448ms/epoch - 9ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0220 - val_loss: 0.0406 - lr: 1.0000e-05 - 418ms/epoch - 9ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0221 - val_loss: 0.0400 - lr: 1.0000e-05 - 402ms/epoch - 8ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0266 - val_loss: 0.0399 - lr: 1.0000e-05 - 421ms/epoch - 9ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0227 - val_loss: 0.0402 - lr: 1.0000e-05 - 448ms/epoch - 9ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0222 - val_loss: 0.0398 - lr: 1.0000e-05 - 452ms/epoch - 9ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0200 - val_loss: 0.0395 - lr: 1.0000e-05 - 431ms/epoch - 9ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0247 - val_loss: 0.0391 - lr: 1.0000e-05 - 424ms/epoch - 9ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0231 - val_loss: 0.0380 - lr: 1.0000e-05 - 473ms/epoch - 10ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0209 - val_loss: 0.0377 - lr: 1.0000e-05 - 465ms/epoch - 10ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0230 - val_loss: 0.0366 - lr: 1.0000e-05 - 432ms/epoch - 9ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0243 - val_loss: 0.0358 - lr: 1.0000e-05 - 427ms/epoch - 9ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0213 - val_loss: 0.0364 - lr: 1.0000e-05 - 419ms/epoch - 9ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0240 - val_loss: 0.0372 - lr: 1.0000e-05 - 383ms/epoch - 8ms/step Epoch 53/500 Epoch 00053: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0247 - val_loss: 0.0364 - lr: 1.0000e-05 - 374ms/epoch - 8ms/step Epoch 54/500 Epoch 00054: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0224 - val_loss: 0.0363 - lr: 1.0000e-05 - 401ms/epoch - 8ms/step Epoch 55/500 Epoch 00055: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0240 - val_loss: 0.0361 - lr: 1.0000e-05 - 431ms/epoch - 9ms/step Epoch 56/500 Epoch 00056: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0228 - val_loss: 0.0369 - lr: 1.0000e-05 - 410ms/epoch - 9ms/step Epoch 57/500 Epoch 00057: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0221 - val_loss: 0.0372 - lr: 1.0000e-05 - 399ms/epoch - 8ms/step Epoch 58/500 Epoch 00058: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0231 - val_loss: 0.0376 - lr: 1.0000e-05 - 407ms/epoch - 8ms/step Epoch 59/500 Epoch 00059: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0218 - val_loss: 0.0367 - lr: 1.0000e-05 - 468ms/epoch - 10ms/step Epoch 60/500 Epoch 00060: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0215 - val_loss: 0.0360 - lr: 1.0000e-05 - 389ms/epoch - 8ms/step Epoch 61/500 Epoch 00061: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0198 - val_loss: 0.0352 - lr: 1.0000e-05 - 461ms/epoch - 10ms/step Epoch 62/500 Epoch 00062: val_loss did not improve from 0.00495 48/48 - 0s - loss: 0.0232 - val_loss: 0.0353 - lr: 1.0000e-05 - 457ms/epoch - 10ms/step Epoch 00062: early stopping
SMA
Prediction vs Close: 51.87% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 39.29635655476263
RMSE: 6.268680607174258
MAPE: 5.07416987111494
EMA
EMA([input_arrays], [timeperiod=30])
Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
51
Working on EMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.41 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4231.556, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3761.238, Time=0.05 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.27 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3532.227, Time=0.08 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3394.496, Time=0.10 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=0.91 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.65 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3396.496, Time=0.20 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 2.699 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1693.248
Date: Sun, 12 Dec 2021 AIC 3394.496
Time: 13:23:46 BIC 3413.260
Sample: 0 HQIC 3401.702
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1982 0.003 -389.569 0.000 -1.204 -1.192
ar.L2 -0.8976 0.006 -139.811 0.000 -0.910 -0.885
ar.L3 -0.3984 0.006 -68.662 0.000 -0.410 -0.387
sigma2 3.9230 0.018 215.372 0.000 3.887 3.959
===================================================================================
Ljung-Box (L1) (Q): 14.54 Jarque-Bera (JB): 2462173.05
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.90
Prob(H) (two-sided): 0.00 Kurtosis: 273.82
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
WARNING:tensorflow:Layer lstm_1 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
WARNING:tensorflow:Layer lstm_1 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.25069, saving model to LSTM1.h5 16/16 - 2s - loss: 0.3653 - val_loss: 0.2507 - lr: 0.0010 - 2s/epoch - 106ms/step Epoch 2/500 Epoch 00002: val_loss improved from 0.25069 to 0.02507, saving model to LSTM1.h5 16/16 - 0s - loss: 0.1125 - val_loss: 0.0251 - lr: 0.0010 - 163ms/epoch - 10ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0745 - val_loss: 0.2931 - lr: 0.0010 - 149ms/epoch - 9ms/step Epoch 4/500 Epoch 00004: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0906 - val_loss: 0.0958 - lr: 0.0010 - 135ms/epoch - 8ms/step Epoch 5/500 Epoch 00005: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0673 - val_loss: 0.0462 - lr: 0.0010 - 131ms/epoch - 8ms/step Epoch 6/500 Epoch 00006: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0710 - val_loss: 0.0599 - lr: 0.0010 - 149ms/epoch - 9ms/step Epoch 7/500 Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00007: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0626 - val_loss: 0.0678 - lr: 0.0010 - 179ms/epoch - 11ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0406 - val_loss: 0.0662 - lr: 1.0000e-04 - 162ms/epoch - 10ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0345 - val_loss: 0.0645 - lr: 1.0000e-04 - 146ms/epoch - 9ms/step Epoch 10/500 Epoch 00010: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0382 - val_loss: 0.0630 - lr: 1.0000e-04 - 147ms/epoch - 9ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0375 - val_loss: 0.0592 - lr: 1.0000e-04 - 139ms/epoch - 9ms/step Epoch 12/500 Epoch 00012: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00012: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0365 - val_loss: 0.0586 - lr: 1.0000e-04 - 153ms/epoch - 10ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0358 - val_loss: 0.0584 - lr: 1.0000e-05 - 148ms/epoch - 9ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0351 - val_loss: 0.0587 - lr: 1.0000e-05 - 157ms/epoch - 10ms/step Epoch 15/500 Epoch 00015: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0382 - val_loss: 0.0586 - lr: 1.0000e-05 - 146ms/epoch - 9ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0378 - val_loss: 0.0588 - lr: 1.0000e-05 - 166ms/epoch - 10ms/step Epoch 17/500 Epoch 00017: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00017: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0364 - val_loss: 0.0587 - lr: 1.0000e-05 - 156ms/epoch - 10ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0381 - val_loss: 0.0587 - lr: 1.0000e-05 - 154ms/epoch - 10ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0355 - val_loss: 0.0590 - lr: 1.0000e-05 - 135ms/epoch - 8ms/step Epoch 20/500 Epoch 00020: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0393 - val_loss: 0.0593 - lr: 1.0000e-05 - 140ms/epoch - 9ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0379 - val_loss: 0.0590 - lr: 1.0000e-05 - 142ms/epoch - 9ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0364 - val_loss: 0.0585 - lr: 1.0000e-05 - 149ms/epoch - 9ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0404 - val_loss: 0.0582 - lr: 1.0000e-05 - 152ms/epoch - 9ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0334 - val_loss: 0.0584 - lr: 1.0000e-05 - 135ms/epoch - 8ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0350 - val_loss: 0.0590 - lr: 1.0000e-05 - 137ms/epoch - 9ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0367 - val_loss: 0.0591 - lr: 1.0000e-05 - 154ms/epoch - 10ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0345 - val_loss: 0.0596 - lr: 1.0000e-05 - 156ms/epoch - 10ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0343 - val_loss: 0.0593 - lr: 1.0000e-05 - 139ms/epoch - 9ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0348 - val_loss: 0.0592 - lr: 1.0000e-05 - 162ms/epoch - 10ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0341 - val_loss: 0.0594 - lr: 1.0000e-05 - 186ms/epoch - 12ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0367 - val_loss: 0.0587 - lr: 1.0000e-05 - 170ms/epoch - 11ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0366 - val_loss: 0.0584 - lr: 1.0000e-05 - 162ms/epoch - 10ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0354 - val_loss: 0.0588 - lr: 1.0000e-05 - 175ms/epoch - 11ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0366 - val_loss: 0.0595 - lr: 1.0000e-05 - 160ms/epoch - 10ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0356 - val_loss: 0.0602 - lr: 1.0000e-05 - 160ms/epoch - 10ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0364 - val_loss: 0.0604 - lr: 1.0000e-05 - 166ms/epoch - 10ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0348 - val_loss: 0.0601 - lr: 1.0000e-05 - 180ms/epoch - 11ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0331 - val_loss: 0.0599 - lr: 1.0000e-05 - 139ms/epoch - 9ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0327 - val_loss: 0.0597 - lr: 1.0000e-05 - 146ms/epoch - 9ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0378 - val_loss: 0.0594 - lr: 1.0000e-05 - 186ms/epoch - 12ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0351 - val_loss: 0.0592 - lr: 1.0000e-05 - 145ms/epoch - 9ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0361 - val_loss: 0.0587 - lr: 1.0000e-05 - 160ms/epoch - 10ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0376 - val_loss: 0.0588 - lr: 1.0000e-05 - 156ms/epoch - 10ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0362 - val_loss: 0.0587 - lr: 1.0000e-05 - 148ms/epoch - 9ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0341 - val_loss: 0.0581 - lr: 1.0000e-05 - 137ms/epoch - 9ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0380 - val_loss: 0.0580 - lr: 1.0000e-05 - 157ms/epoch - 10ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0401 - val_loss: 0.0582 - lr: 1.0000e-05 - 168ms/epoch - 11ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0366 - val_loss: 0.0581 - lr: 1.0000e-05 - 163ms/epoch - 10ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0348 - val_loss: 0.0580 - lr: 1.0000e-05 - 164ms/epoch - 10ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0344 - val_loss: 0.0578 - lr: 1.0000e-05 - 142ms/epoch - 9ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0339 - val_loss: 0.0575 - lr: 1.0000e-05 - 181ms/epoch - 11ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.02507 16/16 - 0s - loss: 0.0369 - val_loss: 0.0578 - lr: 1.0000e-05 - 155ms/epoch - 10ms/step Epoch 00052: early stopping
SMA
Prediction vs Close: 51.87% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 39.29635655476263
RMSE: 6.268680607174258
MAPE: 5.07416987111494
EMA
Prediction vs Close: 56.34% Accuracy
Prediction vs Prediction: 48.88% Accuracy
MSE: 30.25057164247689
RMSE: 5.50005196725239
MAPE: 4.444486270049439
WMA
WMA([input_arrays], [timeperiod=30])
Weighted Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
49
Working on WMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.41 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4264.089, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3793.930, Time=0.05 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.25 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3564.923, Time=0.07 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3427.258, Time=0.08 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.41 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.52 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3429.258, Time=0.20 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.029 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1709.629
Date: Sun, 12 Dec 2021 AIC 3427.258
Time: 13:25:07 BIC 3446.021
Sample: 0 HQIC 3434.464
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1981 0.003 -389.386 0.000 -1.204 -1.192
ar.L2 -0.8974 0.006 -139.699 0.000 -0.910 -0.885
ar.L3 -0.3983 0.006 -68.737 0.000 -0.410 -0.387
sigma2 4.0860 0.019 215.311 0.000 4.049 4.123
===================================================================================
Ljung-Box (L1) (Q): 14.57 Jarque-Bera (JB): 2460901.70
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.90
Prob(H) (two-sided): 0.00 Kurtosis: 273.75
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
WARNING:tensorflow:Layer lstm_2 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
WARNING:tensorflow:Layer lstm_2 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.07191, saving model to LSTM1.h5 17/17 - 2s - loss: 0.5361 - val_loss: 0.0719 - lr: 0.0010 - 2s/epoch - 121ms/step Epoch 2/500 Epoch 00002: val_loss improved from 0.07191 to 0.04083, saving model to LSTM1.h5 17/17 - 0s - loss: 0.2156 - val_loss: 0.0408 - lr: 0.0010 - 167ms/epoch - 10ms/step Epoch 3/500 Epoch 00003: val_loss improved from 0.04083 to 0.03980, saving model to LSTM1.h5 17/17 - 0s - loss: 0.0759 - val_loss: 0.0398 - lr: 0.0010 - 174ms/epoch - 10ms/step Epoch 4/500 Epoch 00004: val_loss did not improve from 0.03980 17/17 - 0s - loss: 0.0752 - val_loss: 0.0900 - lr: 0.0010 - 197ms/epoch - 12ms/step Epoch 5/500 Epoch 00005: val_loss did not improve from 0.03980 17/17 - 0s - loss: 0.0786 - val_loss: 0.0610 - lr: 0.0010 - 149ms/epoch - 9ms/step Epoch 6/500 Epoch 00006: val_loss improved from 0.03980 to 0.02048, saving model to LSTM1.h5 17/17 - 0s - loss: 0.0640 - val_loss: 0.0205 - lr: 0.0010 - 166ms/epoch - 10ms/step Epoch 7/500 Epoch 00007: val_loss did not improve from 0.02048 17/17 - 0s - loss: 0.0500 - val_loss: 0.3935 - lr: 0.0010 - 151ms/epoch - 9ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.02048 17/17 - 0s - loss: 0.0583 - val_loss: 0.0728 - lr: 0.0010 - 146ms/epoch - 9ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.02048 17/17 - 0s - loss: 0.0530 - val_loss: 0.0487 - lr: 0.0010 - 160ms/epoch - 9ms/step Epoch 10/500 Epoch 00010: val_loss improved from 0.02048 to 0.00685, saving model to LSTM1.h5 17/17 - 0s - loss: 0.0460 - val_loss: 0.0069 - lr: 0.0010 - 176ms/epoch - 10ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0505 - val_loss: 0.1364 - lr: 0.0010 - 144ms/epoch - 8ms/step Epoch 12/500 Epoch 00012: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0399 - val_loss: 0.0105 - lr: 0.0010 - 149ms/epoch - 9ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0350 - val_loss: 0.0142 - lr: 0.0010 - 192ms/epoch - 11ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0362 - val_loss: 0.0711 - lr: 0.0010 - 163ms/epoch - 10ms/step Epoch 15/500 Epoch 00015: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00015: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0354 - val_loss: 0.0117 - lr: 0.0010 - 169ms/epoch - 10ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0376 - val_loss: 0.0131 - lr: 1.0000e-04 - 154ms/epoch - 9ms/step Epoch 17/500 Epoch 00017: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0317 - val_loss: 0.0129 - lr: 1.0000e-04 - 153ms/epoch - 9ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0325 - val_loss: 0.0139 - lr: 1.0000e-04 - 155ms/epoch - 9ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0281 - val_loss: 0.0138 - lr: 1.0000e-04 - 173ms/epoch - 10ms/step Epoch 20/500 Epoch 00020: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00020: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0307 - val_loss: 0.0125 - lr: 1.0000e-04 - 166ms/epoch - 10ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0300 - val_loss: 0.0125 - lr: 1.0000e-05 - 155ms/epoch - 9ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0299 - val_loss: 0.0123 - lr: 1.0000e-05 - 149ms/epoch - 9ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0292 - val_loss: 0.0123 - lr: 1.0000e-05 - 148ms/epoch - 9ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0296 - val_loss: 0.0122 - lr: 1.0000e-05 - 158ms/epoch - 9ms/step Epoch 25/500 Epoch 00025: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00025: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0289 - val_loss: 0.0125 - lr: 1.0000e-05 - 145ms/epoch - 9ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0287 - val_loss: 0.0128 - lr: 1.0000e-05 - 157ms/epoch - 9ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0299 - val_loss: 0.0129 - lr: 1.0000e-05 - 154ms/epoch - 9ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0295 - val_loss: 0.0130 - lr: 1.0000e-05 - 143ms/epoch - 8ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0307 - val_loss: 0.0133 - lr: 1.0000e-05 - 168ms/epoch - 10ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0284 - val_loss: 0.0134 - lr: 1.0000e-05 - 155ms/epoch - 9ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0318 - val_loss: 0.0133 - lr: 1.0000e-05 - 154ms/epoch - 9ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0295 - val_loss: 0.0136 - lr: 1.0000e-05 - 172ms/epoch - 10ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0285 - val_loss: 0.0135 - lr: 1.0000e-05 - 173ms/epoch - 10ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0281 - val_loss: 0.0135 - lr: 1.0000e-05 - 141ms/epoch - 8ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0304 - val_loss: 0.0138 - lr: 1.0000e-05 - 148ms/epoch - 9ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0304 - val_loss: 0.0141 - lr: 1.0000e-05 - 145ms/epoch - 9ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0303 - val_loss: 0.0139 - lr: 1.0000e-05 - 155ms/epoch - 9ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0282 - val_loss: 0.0138 - lr: 1.0000e-05 - 153ms/epoch - 9ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0268 - val_loss: 0.0138 - lr: 1.0000e-05 - 148ms/epoch - 9ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0314 - val_loss: 0.0142 - lr: 1.0000e-05 - 152ms/epoch - 9ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0280 - val_loss: 0.0142 - lr: 1.0000e-05 - 140ms/epoch - 8ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0309 - val_loss: 0.0141 - lr: 1.0000e-05 - 152ms/epoch - 9ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0304 - val_loss: 0.0144 - lr: 1.0000e-05 - 158ms/epoch - 9ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0295 - val_loss: 0.0145 - lr: 1.0000e-05 - 143ms/epoch - 8ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0296 - val_loss: 0.0147 - lr: 1.0000e-05 - 166ms/epoch - 10ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0330 - val_loss: 0.0146 - lr: 1.0000e-05 - 157ms/epoch - 9ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0286 - val_loss: 0.0146 - lr: 1.0000e-05 - 156ms/epoch - 9ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0307 - val_loss: 0.0142 - lr: 1.0000e-05 - 169ms/epoch - 10ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0280 - val_loss: 0.0140 - lr: 1.0000e-05 - 153ms/epoch - 9ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0290 - val_loss: 0.0138 - lr: 1.0000e-05 - 154ms/epoch - 9ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0292 - val_loss: 0.0137 - lr: 1.0000e-05 - 161ms/epoch - 9ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0308 - val_loss: 0.0135 - lr: 1.0000e-05 - 151ms/epoch - 9ms/step Epoch 53/500 Epoch 00053: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0303 - val_loss: 0.0138 - lr: 1.0000e-05 - 157ms/epoch - 9ms/step Epoch 54/500 Epoch 00054: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0301 - val_loss: 0.0140 - lr: 1.0000e-05 - 158ms/epoch - 9ms/step Epoch 55/500 Epoch 00055: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0299 - val_loss: 0.0140 - lr: 1.0000e-05 - 166ms/epoch - 10ms/step Epoch 56/500 Epoch 00056: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0295 - val_loss: 0.0140 - lr: 1.0000e-05 - 149ms/epoch - 9ms/step Epoch 57/500 Epoch 00057: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0293 - val_loss: 0.0138 - lr: 1.0000e-05 - 155ms/epoch - 9ms/step Epoch 58/500 Epoch 00058: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0301 - val_loss: 0.0134 - lr: 1.0000e-05 - 158ms/epoch - 9ms/step Epoch 59/500 Epoch 00059: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0342 - val_loss: 0.0134 - lr: 1.0000e-05 - 152ms/epoch - 9ms/step Epoch 60/500 Epoch 00060: val_loss did not improve from 0.00685 17/17 - 0s - loss: 0.0271 - val_loss: 0.0138 - lr: 1.0000e-05 - 171ms/epoch - 10ms/step Epoch 00060: early stopping
SMA
Prediction vs Close: 51.87% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 39.29635655476263
RMSE: 6.268680607174258
MAPE: 5.07416987111494
EMA
Prediction vs Close: 56.34% Accuracy
Prediction vs Prediction: 48.88% Accuracy
MSE: 30.25057164247689
RMSE: 5.50005196725239
MAPE: 4.444486270049439
WMA
Prediction vs Close: 57.84% Accuracy
Prediction vs Prediction: 41.79% Accuracy
MSE: 64.92643731055051
RMSE: 8.057694292448089
MAPE: 6.289003841800032
DEMA
DEMA([input_arrays], [timeperiod=30])
Double Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
89
Working on DEMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.39 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4436.126, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3965.317, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.38 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3736.589, Time=0.07 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3598.951, Time=0.08 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.00 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.95 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3600.951, Time=0.18 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.122 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1795.475
Date: Sun, 12 Dec 2021 AIC 3598.951
Time: 13:26:28 BIC 3617.714
Sample: 0 HQIC 3606.157
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1983 0.003 -389.581 0.000 -1.204 -1.192
ar.L2 -0.8973 0.006 -139.732 0.000 -0.910 -0.885
ar.L3 -0.3983 0.006 -68.649 0.000 -0.410 -0.387
sigma2 5.0573 0.023 215.292 0.000 5.011 5.103
===================================================================================
Ljung-Box (L1) (Q): 14.41 Jarque-Bera (JB): 2460553.80
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.89
Prob(H) (two-sided): 0.00 Kurtosis: 273.74
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
WARNING:tensorflow:Layer lstm_3 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
WARNING:tensorflow:Layer lstm_3 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
Epoch 1/500 Epoch 00001: val_loss improved from inf to 1.03480, saving model to LSTM1.h5 10/10 - 2s - loss: 0.2906 - val_loss: 1.0348 - lr: 0.0010 - 2s/epoch - 166ms/step Epoch 2/500 Epoch 00002: val_loss improved from 1.03480 to 0.10184, saving model to LSTM1.h5 10/10 - 0s - loss: 0.0729 - val_loss: 0.1018 - lr: 0.0010 - 145ms/epoch - 15ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0690 - val_loss: 0.3825 - lr: 0.0010 - 109ms/epoch - 11ms/step Epoch 4/500 Epoch 00004: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0559 - val_loss: 0.3389 - lr: 0.0010 - 112ms/epoch - 11ms/step Epoch 5/500 Epoch 00005: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0561 - val_loss: 0.1873 - lr: 0.0010 - 105ms/epoch - 11ms/step Epoch 6/500 Epoch 00006: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0567 - val_loss: 0.1386 - lr: 0.0010 - 101ms/epoch - 10ms/step Epoch 7/500 Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00007: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0669 - val_loss: 0.1327 - lr: 0.0010 - 105ms/epoch - 11ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0567 - val_loss: 0.1416 - lr: 1.0000e-04 - 120ms/epoch - 12ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0443 - val_loss: 0.1491 - lr: 1.0000e-04 - 117ms/epoch - 12ms/step Epoch 10/500 Epoch 00010: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0430 - val_loss: 0.1551 - lr: 1.0000e-04 - 96ms/epoch - 10ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0399 - val_loss: 0.1599 - lr: 1.0000e-04 - 90ms/epoch - 9ms/step Epoch 12/500 Epoch 00012: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00012: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0371 - val_loss: 0.1642 - lr: 1.0000e-04 - 106ms/epoch - 11ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0344 - val_loss: 0.1644 - lr: 1.0000e-05 - 141ms/epoch - 14ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0371 - val_loss: 0.1646 - lr: 1.0000e-05 - 119ms/epoch - 12ms/step Epoch 15/500 Epoch 00015: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0331 - val_loss: 0.1650 - lr: 1.0000e-05 - 128ms/epoch - 13ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0365 - val_loss: 0.1656 - lr: 1.0000e-05 - 115ms/epoch - 12ms/step Epoch 17/500 Epoch 00017: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00017: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0348 - val_loss: 0.1660 - lr: 1.0000e-05 - 117ms/epoch - 12ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0391 - val_loss: 0.1664 - lr: 1.0000e-05 - 105ms/epoch - 11ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0390 - val_loss: 0.1670 - lr: 1.0000e-05 - 100ms/epoch - 10ms/step Epoch 20/500 Epoch 00020: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0352 - val_loss: 0.1675 - lr: 1.0000e-05 - 107ms/epoch - 11ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0370 - val_loss: 0.1678 - lr: 1.0000e-05 - 102ms/epoch - 10ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0318 - val_loss: 0.1682 - lr: 1.0000e-05 - 114ms/epoch - 11ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0363 - val_loss: 0.1688 - lr: 1.0000e-05 - 118ms/epoch - 12ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0355 - val_loss: 0.1689 - lr: 1.0000e-05 - 96ms/epoch - 10ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0321 - val_loss: 0.1695 - lr: 1.0000e-05 - 106ms/epoch - 11ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0345 - val_loss: 0.1698 - lr: 1.0000e-05 - 105ms/epoch - 10ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0359 - val_loss: 0.1701 - lr: 1.0000e-05 - 105ms/epoch - 10ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0377 - val_loss: 0.1702 - lr: 1.0000e-05 - 131ms/epoch - 13ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0347 - val_loss: 0.1702 - lr: 1.0000e-05 - 123ms/epoch - 12ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0357 - val_loss: 0.1700 - lr: 1.0000e-05 - 115ms/epoch - 11ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0318 - val_loss: 0.1699 - lr: 1.0000e-05 - 108ms/epoch - 11ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0379 - val_loss: 0.1708 - lr: 1.0000e-05 - 107ms/epoch - 11ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0378 - val_loss: 0.1723 - lr: 1.0000e-05 - 117ms/epoch - 12ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0353 - val_loss: 0.1733 - lr: 1.0000e-05 - 106ms/epoch - 11ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0340 - val_loss: 0.1732 - lr: 1.0000e-05 - 105ms/epoch - 10ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0321 - val_loss: 0.1733 - lr: 1.0000e-05 - 103ms/epoch - 10ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0382 - val_loss: 0.1737 - lr: 1.0000e-05 - 89ms/epoch - 9ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0360 - val_loss: 0.1739 - lr: 1.0000e-05 - 96ms/epoch - 10ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0353 - val_loss: 0.1743 - lr: 1.0000e-05 - 95ms/epoch - 10ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0308 - val_loss: 0.1740 - lr: 1.0000e-05 - 115ms/epoch - 11ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0347 - val_loss: 0.1742 - lr: 1.0000e-05 - 105ms/epoch - 11ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0338 - val_loss: 0.1743 - lr: 1.0000e-05 - 101ms/epoch - 10ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0333 - val_loss: 0.1738 - lr: 1.0000e-05 - 103ms/epoch - 10ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0344 - val_loss: 0.1733 - lr: 1.0000e-05 - 117ms/epoch - 12ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0357 - val_loss: 0.1731 - lr: 1.0000e-05 - 95ms/epoch - 10ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0369 - val_loss: 0.1735 - lr: 1.0000e-05 - 122ms/epoch - 12ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0343 - val_loss: 0.1740 - lr: 1.0000e-05 - 99ms/epoch - 10ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0378 - val_loss: 0.1753 - lr: 1.0000e-05 - 92ms/epoch - 9ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0347 - val_loss: 0.1759 - lr: 1.0000e-05 - 88ms/epoch - 9ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0349 - val_loss: 0.1764 - lr: 1.0000e-05 - 115ms/epoch - 11ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0360 - val_loss: 0.1769 - lr: 1.0000e-05 - 107ms/epoch - 11ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.10184 10/10 - 0s - loss: 0.0301 - val_loss: 0.1769 - lr: 1.0000e-05 - 109ms/epoch - 11ms/step Epoch 00052: early stopping
SMA
Prediction vs Close: 51.87% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 39.29635655476263
RMSE: 6.268680607174258
MAPE: 5.07416987111494
EMA
Prediction vs Close: 56.34% Accuracy
Prediction vs Prediction: 48.88% Accuracy
MSE: 30.25057164247689
RMSE: 5.50005196725239
MAPE: 4.444486270049439
WMA
Prediction vs Close: 57.84% Accuracy
Prediction vs Prediction: 41.79% Accuracy
MSE: 64.92643731055051
RMSE: 8.057694292448089
MAPE: 6.289003841800032
DEMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 46.27% Accuracy
MSE: 52.98651621196421
RMSE: 7.279183760008
MAPE: 5.725540843661134
KAMA
KAMA([input_arrays], [timeperiod=30])
Kaufman Adaptive Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
18
Working on KAMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.38 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4190.464, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3724.371, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.28 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3494.154, Time=0.07 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3357.435, Time=0.09 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.23 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.75 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3359.435, Time=0.20 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.079 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1674.717
Date: Sun, 12 Dec 2021 AIC 3357.435
Time: 13:27:39 BIC 3376.198
Sample: 0 HQIC 3364.641
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1955 0.003 -381.246 0.000 -1.202 -1.189
ar.L2 -0.8964 0.007 -135.835 0.000 -0.909 -0.883
ar.L3 -0.3971 0.006 -67.229 0.000 -0.409 -0.385
sigma2 3.7466 0.018 211.623 0.000 3.712 3.781
===================================================================================
Ljung-Box (L1) (Q): 14.20 Jarque-Bera (JB): 2338363.32
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.01 Skew: 3.76
Prob(H) (two-sided): 0.00 Kurtosis: 266.93
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
WARNING:tensorflow:Layer lstm_4 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
WARNING:tensorflow:Layer lstm_4 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.15184, saving model to LSTM1.h5 45/45 - 2s - loss: 0.3564 - val_loss: 0.1518 - lr: 0.0010 - 2s/epoch - 43ms/step Epoch 2/500 Epoch 00002: val_loss did not improve from 0.15184 45/45 - 0s - loss: 0.1500 - val_loss: 0.3345 - lr: 0.0010 - 401ms/epoch - 9ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.15184 45/45 - 0s - loss: 0.0501 - val_loss: 0.7766 - lr: 0.0010 - 368ms/epoch - 8ms/step Epoch 4/500 Epoch 00004: val_loss improved from 0.15184 to 0.09772, saving model to LSTM1.h5 45/45 - 0s - loss: 0.0492 - val_loss: 0.0977 - lr: 0.0010 - 437ms/epoch - 10ms/step Epoch 5/500 Epoch 00005: val_loss improved from 0.09772 to 0.01025, saving model to LSTM1.h5 45/45 - 0s - loss: 0.0431 - val_loss: 0.0102 - lr: 0.0010 - 464ms/epoch - 10ms/step Epoch 6/500 Epoch 00006: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0480 - val_loss: 0.1955 - lr: 0.0010 - 408ms/epoch - 9ms/step Epoch 7/500 Epoch 00007: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0356 - val_loss: 0.0301 - lr: 0.0010 - 414ms/epoch - 9ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0279 - val_loss: 0.0184 - lr: 0.0010 - 402ms/epoch - 9ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0293 - val_loss: 0.0412 - lr: 0.0010 - 365ms/epoch - 8ms/step Epoch 10/500 Epoch 00010: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00010: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0301 - val_loss: 0.0185 - lr: 0.0010 - 433ms/epoch - 10ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0314 - val_loss: 0.0292 - lr: 1.0000e-04 - 451ms/epoch - 10ms/step Epoch 12/500 Epoch 00012: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0258 - val_loss: 0.0291 - lr: 1.0000e-04 - 415ms/epoch - 9ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0258 - val_loss: 0.0261 - lr: 1.0000e-04 - 426ms/epoch - 9ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0278 - val_loss: 0.0284 - lr: 1.0000e-04 - 361ms/epoch - 8ms/step Epoch 15/500 Epoch 00015: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00015: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0258 - val_loss: 0.0303 - lr: 1.0000e-04 - 388ms/epoch - 9ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0223 - val_loss: 0.0307 - lr: 1.0000e-05 - 410ms/epoch - 9ms/step Epoch 17/500 Epoch 00017: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0262 - val_loss: 0.0304 - lr: 1.0000e-05 - 391ms/epoch - 9ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0246 - val_loss: 0.0309 - lr: 1.0000e-05 - 388ms/epoch - 9ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0240 - val_loss: 0.0315 - lr: 1.0000e-05 - 395ms/epoch - 9ms/step Epoch 20/500 Epoch 00020: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00020: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0226 - val_loss: 0.0319 - lr: 1.0000e-05 - 395ms/epoch - 9ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0244 - val_loss: 0.0322 - lr: 1.0000e-05 - 399ms/epoch - 9ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0234 - val_loss: 0.0318 - lr: 1.0000e-05 - 448ms/epoch - 10ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0248 - val_loss: 0.0319 - lr: 1.0000e-05 - 405ms/epoch - 9ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0235 - val_loss: 0.0318 - lr: 1.0000e-05 - 391ms/epoch - 9ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0225 - val_loss: 0.0322 - lr: 1.0000e-05 - 350ms/epoch - 8ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0240 - val_loss: 0.0322 - lr: 1.0000e-05 - 401ms/epoch - 9ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0250 - val_loss: 0.0323 - lr: 1.0000e-05 - 407ms/epoch - 9ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0247 - val_loss: 0.0330 - lr: 1.0000e-05 - 381ms/epoch - 8ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0237 - val_loss: 0.0336 - lr: 1.0000e-05 - 375ms/epoch - 8ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0235 - val_loss: 0.0340 - lr: 1.0000e-05 - 359ms/epoch - 8ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0223 - val_loss: 0.0341 - lr: 1.0000e-05 - 429ms/epoch - 10ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0214 - val_loss: 0.0346 - lr: 1.0000e-05 - 430ms/epoch - 10ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0241 - val_loss: 0.0347 - lr: 1.0000e-05 - 399ms/epoch - 9ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0220 - val_loss: 0.0359 - lr: 1.0000e-05 - 392ms/epoch - 9ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0217 - val_loss: 0.0356 - lr: 1.0000e-05 - 389ms/epoch - 9ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0232 - val_loss: 0.0345 - lr: 1.0000e-05 - 381ms/epoch - 8ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0248 - val_loss: 0.0340 - lr: 1.0000e-05 - 372ms/epoch - 8ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0236 - val_loss: 0.0344 - lr: 1.0000e-05 - 391ms/epoch - 9ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0232 - val_loss: 0.0348 - lr: 1.0000e-05 - 401ms/epoch - 9ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0234 - val_loss: 0.0341 - lr: 1.0000e-05 - 376ms/epoch - 8ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0204 - val_loss: 0.0340 - lr: 1.0000e-05 - 388ms/epoch - 9ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0245 - val_loss: 0.0349 - lr: 1.0000e-05 - 385ms/epoch - 9ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0219 - val_loss: 0.0355 - lr: 1.0000e-05 - 441ms/epoch - 10ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0220 - val_loss: 0.0353 - lr: 1.0000e-05 - 413ms/epoch - 9ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0220 - val_loss: 0.0357 - lr: 1.0000e-05 - 359ms/epoch - 8ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0219 - val_loss: 0.0367 - lr: 1.0000e-05 - 388ms/epoch - 9ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0220 - val_loss: 0.0361 - lr: 1.0000e-05 - 421ms/epoch - 9ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0236 - val_loss: 0.0353 - lr: 1.0000e-05 - 419ms/epoch - 9ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0222 - val_loss: 0.0359 - lr: 1.0000e-05 - 493ms/epoch - 11ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0249 - val_loss: 0.0355 - lr: 1.0000e-05 - 470ms/epoch - 10ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0228 - val_loss: 0.0357 - lr: 1.0000e-05 - 383ms/epoch - 9ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0229 - val_loss: 0.0348 - lr: 1.0000e-05 - 408ms/epoch - 9ms/step Epoch 53/500 Epoch 00053: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0229 - val_loss: 0.0345 - lr: 1.0000e-05 - 393ms/epoch - 9ms/step Epoch 54/500 Epoch 00054: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0221 - val_loss: 0.0336 - lr: 1.0000e-05 - 469ms/epoch - 10ms/step Epoch 55/500 Epoch 00055: val_loss did not improve from 0.01025 45/45 - 0s - loss: 0.0253 - val_loss: 0.0328 - lr: 1.0000e-05 - 427ms/epoch - 9ms/step Epoch 00055: early stopping
SMA
Prediction vs Close: 51.87% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 39.29635655476263
RMSE: 6.268680607174258
MAPE: 5.07416987111494
EMA
Prediction vs Close: 56.34% Accuracy
Prediction vs Prediction: 48.88% Accuracy
MSE: 30.25057164247689
RMSE: 5.50005196725239
MAPE: 4.444486270049439
WMA
Prediction vs Close: 57.84% Accuracy
Prediction vs Prediction: 41.79% Accuracy
MSE: 64.92643731055051
RMSE: 8.057694292448089
MAPE: 6.289003841800032
DEMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 46.27% Accuracy
MSE: 52.98651621196421
RMSE: 7.279183760008
MAPE: 5.725540843661134
KAMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 33.6026987861483
RMSE: 5.796783486223054
MAPE: 4.518981487962124
MIDPOINT
MIDPOINT([input_arrays], [timeperiod=14])
MidPoint over period (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 14
Outputs:
real
14
Working on MIDPOINT predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.35 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4212.289, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3747.746, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.27 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3523.401, Time=0.10 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3387.759, Time=0.12 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.42 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=1.00 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3389.758, Time=0.20 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.562 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1689.879
Date: Sun, 12 Dec 2021 AIC 3387.759
Time: 13:29:32 BIC 3406.522
Sample: 0 HQIC 3394.964
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1878 0.003 -345.315 0.000 -1.195 -1.181
ar.L2 -0.8876 0.007 -121.809 0.000 -0.902 -0.873
ar.L3 -0.3957 0.007 -60.127 0.000 -0.409 -0.383
sigma2 3.8904 0.020 193.404 0.000 3.851 3.930
===================================================================================
Ljung-Box (L1) (Q): 13.21 Jarque-Bera (JB): 1659080.01
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.08 Skew: 3.28
Prob(H) (two-sided): 0.00 Kurtosis: 225.31
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
WARNING:tensorflow:Layer lstm_5 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
WARNING:tensorflow:Layer lstm_5 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.02821, saving model to LSTM1.h5 58/58 - 2s - loss: 0.1914 - val_loss: 0.0282 - lr: 0.0010 - 2s/epoch - 40ms/step Epoch 2/500 Epoch 00002: val_loss improved from 0.02821 to 0.02136, saving model to LSTM1.h5 58/58 - 1s - loss: 0.0663 - val_loss: 0.0214 - lr: 0.0010 - 502ms/epoch - 9ms/step Epoch 3/500 Epoch 00003: val_loss improved from 0.02136 to 0.00821, saving model to LSTM1.h5 58/58 - 1s - loss: 0.1170 - val_loss: 0.0082 - lr: 0.0010 - 514ms/epoch - 9ms/step Epoch 4/500 Epoch 00004: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0619 - val_loss: 0.5512 - lr: 0.0010 - 496ms/epoch - 9ms/step Epoch 5/500 Epoch 00005: val_loss did not improve from 0.00821 58/58 - 1s - loss: 0.0423 - val_loss: 0.4569 - lr: 0.0010 - 519ms/epoch - 9ms/step Epoch 6/500 Epoch 00006: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0480 - val_loss: 0.2682 - lr: 0.0010 - 491ms/epoch - 8ms/step Epoch 7/500 Epoch 00007: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0383 - val_loss: 0.0790 - lr: 0.0010 - 483ms/epoch - 8ms/step Epoch 8/500 Epoch 00008: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00008: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0367 - val_loss: 0.2302 - lr: 0.0010 - 494ms/epoch - 9ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0332 - val_loss: 0.2024 - lr: 1.0000e-04 - 449ms/epoch - 8ms/step Epoch 10/500 Epoch 00010: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0316 - val_loss: 0.1622 - lr: 1.0000e-04 - 484ms/epoch - 8ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0313 - val_loss: 0.1290 - lr: 1.0000e-04 - 497ms/epoch - 9ms/step Epoch 12/500 Epoch 00012: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0296 - val_loss: 0.1063 - lr: 1.0000e-04 - 465ms/epoch - 8ms/step Epoch 13/500 Epoch 00013: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00013: val_loss did not improve from 0.00821 58/58 - 1s - loss: 0.0318 - val_loss: 0.0985 - lr: 1.0000e-04 - 521ms/epoch - 9ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0288 - val_loss: 0.0967 - lr: 1.0000e-05 - 471ms/epoch - 8ms/step Epoch 15/500 Epoch 00015: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0306 - val_loss: 0.0950 - lr: 1.0000e-05 - 482ms/epoch - 8ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0295 - val_loss: 0.0936 - lr: 1.0000e-05 - 486ms/epoch - 8ms/step Epoch 17/500 Epoch 00017: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0285 - val_loss: 0.0924 - lr: 1.0000e-05 - 469ms/epoch - 8ms/step Epoch 18/500 Epoch 00018: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00018: val_loss did not improve from 0.00821 58/58 - 1s - loss: 0.0288 - val_loss: 0.0918 - lr: 1.0000e-05 - 522ms/epoch - 9ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.00821 58/58 - 1s - loss: 0.0278 - val_loss: 0.0913 - lr: 1.0000e-05 - 500ms/epoch - 9ms/step Epoch 20/500 Epoch 00020: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0280 - val_loss: 0.0904 - lr: 1.0000e-05 - 479ms/epoch - 8ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.00821 58/58 - 1s - loss: 0.0282 - val_loss: 0.0887 - lr: 1.0000e-05 - 509ms/epoch - 9ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0291 - val_loss: 0.0879 - lr: 1.0000e-05 - 496ms/epoch - 9ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.00821 58/58 - 1s - loss: 0.0282 - val_loss: 0.0867 - lr: 1.0000e-05 - 502ms/epoch - 9ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0297 - val_loss: 0.0863 - lr: 1.0000e-05 - 484ms/epoch - 8ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.00821 58/58 - 1s - loss: 0.0264 - val_loss: 0.0848 - lr: 1.0000e-05 - 512ms/epoch - 9ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0282 - val_loss: 0.0843 - lr: 1.0000e-05 - 482ms/epoch - 8ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0270 - val_loss: 0.0849 - lr: 1.0000e-05 - 452ms/epoch - 8ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0292 - val_loss: 0.0852 - lr: 1.0000e-05 - 456ms/epoch - 8ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0273 - val_loss: 0.0833 - lr: 1.0000e-05 - 461ms/epoch - 8ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0299 - val_loss: 0.0808 - lr: 1.0000e-05 - 490ms/epoch - 8ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.00821 58/58 - 1s - loss: 0.0283 - val_loss: 0.0796 - lr: 1.0000e-05 - 526ms/epoch - 9ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.00821 58/58 - 1s - loss: 0.0271 - val_loss: 0.0778 - lr: 1.0000e-05 - 502ms/epoch - 9ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.00821 58/58 - 1s - loss: 0.0291 - val_loss: 0.0749 - lr: 1.0000e-05 - 503ms/epoch - 9ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0270 - val_loss: 0.0752 - lr: 1.0000e-05 - 480ms/epoch - 8ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0268 - val_loss: 0.0763 - lr: 1.0000e-05 - 493ms/epoch - 9ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0271 - val_loss: 0.0781 - lr: 1.0000e-05 - 436ms/epoch - 8ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.00821 58/58 - 1s - loss: 0.0276 - val_loss: 0.0782 - lr: 1.0000e-05 - 524ms/epoch - 9ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0273 - val_loss: 0.0801 - lr: 1.0000e-05 - 488ms/epoch - 8ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0295 - val_loss: 0.0792 - lr: 1.0000e-05 - 499ms/epoch - 9ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0288 - val_loss: 0.0789 - lr: 1.0000e-05 - 478ms/epoch - 8ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.00821 58/58 - 1s - loss: 0.0294 - val_loss: 0.0793 - lr: 1.0000e-05 - 503ms/epoch - 9ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0272 - val_loss: 0.0760 - lr: 1.0000e-05 - 462ms/epoch - 8ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0283 - val_loss: 0.0758 - lr: 1.0000e-05 - 484ms/epoch - 8ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0253 - val_loss: 0.0797 - lr: 1.0000e-05 - 449ms/epoch - 8ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.00821 58/58 - 1s - loss: 0.0293 - val_loss: 0.0781 - lr: 1.0000e-05 - 524ms/epoch - 9ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0254 - val_loss: 0.0773 - lr: 1.0000e-05 - 467ms/epoch - 8ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0271 - val_loss: 0.0796 - lr: 1.0000e-05 - 492ms/epoch - 8ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0275 - val_loss: 0.0797 - lr: 1.0000e-05 - 477ms/epoch - 8ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0275 - val_loss: 0.0832 - lr: 1.0000e-05 - 484ms/epoch - 8ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.00821 58/58 - 1s - loss: 0.0309 - val_loss: 0.0801 - lr: 1.0000e-05 - 522ms/epoch - 9ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0292 - val_loss: 0.0798 - lr: 1.0000e-05 - 481ms/epoch - 8ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0282 - val_loss: 0.0797 - lr: 1.0000e-05 - 456ms/epoch - 8ms/step Epoch 53/500 Epoch 00053: val_loss did not improve from 0.00821 58/58 - 0s - loss: 0.0260 - val_loss: 0.0784 - lr: 1.0000e-05 - 488ms/epoch - 8ms/step Epoch 00053: early stopping
SMA
Prediction vs Close: 51.87% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 39.29635655476263
RMSE: 6.268680607174258
MAPE: 5.07416987111494
EMA
Prediction vs Close: 56.34% Accuracy
Prediction vs Prediction: 48.88% Accuracy
MSE: 30.25057164247689
RMSE: 5.50005196725239
MAPE: 4.444486270049439
WMA
Prediction vs Close: 57.84% Accuracy
Prediction vs Prediction: 41.79% Accuracy
MSE: 64.92643731055051
RMSE: 8.057694292448089
MAPE: 6.289003841800032
DEMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 46.27% Accuracy
MSE: 52.98651621196421
RMSE: 7.279183760008
MAPE: 5.725540843661134
KAMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 33.6026987861483
RMSE: 5.796783486223054
MAPE: 4.518981487962124
MIDPOINT
Prediction vs Close: 50.0% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 22.7414116550439
RMSE: 4.768795618921396
MAPE: 3.944458615157319
T3
T3([input_arrays], [timeperiod=5], [vfactor=0.7])
Triple Exponential Moving Average (T3) (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 5
vfactor: 0.7
Outputs:
real
19
Working on T3 predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.36 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4414.515, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3944.062, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.38 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3715.173, Time=0.08 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3577.471, Time=0.09 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.45 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.62 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3579.471, Time=0.19 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.248 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1784.736
Date: Sun, 12 Dec 2021 AIC 3577.471
Time: 13:31:25 BIC 3596.235
Sample: 0 HQIC 3584.677
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1982 0.003 -389.844 0.000 -1.204 -1.192
ar.L2 -0.8974 0.006 -139.861 0.000 -0.910 -0.885
ar.L3 -0.3983 0.006 -68.862 0.000 -0.410 -0.387
sigma2 4.9242 0.023 215.469 0.000 4.879 4.969
===================================================================================
Ljung-Box (L1) (Q): 14.55 Jarque-Bera (JB): 2468024.38
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.90
Prob(H) (two-sided): 0.00 Kurtosis: 274.15
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
WARNING:tensorflow:Layer lstm_6 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
WARNING:tensorflow:Layer lstm_6 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.05392, saving model to LSTM1.h5 43/43 - 2s - loss: 0.3047 - val_loss: 0.0539 - lr: 0.0010 - 2s/epoch - 44ms/step Epoch 2/500 Epoch 00002: val_loss improved from 0.05392 to 0.04661, saving model to LSTM1.h5 43/43 - 0s - loss: 0.1255 - val_loss: 0.0466 - lr: 0.0010 - 395ms/epoch - 9ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.1140 - val_loss: 0.6637 - lr: 0.0010 - 405ms/epoch - 9ms/step Epoch 4/500 Epoch 00004: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0770 - val_loss: 0.3570 - lr: 0.0010 - 367ms/epoch - 9ms/step Epoch 5/500 Epoch 00005: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0437 - val_loss: 0.1212 - lr: 0.0010 - 431ms/epoch - 10ms/step Epoch 6/500 Epoch 00006: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0622 - val_loss: 0.1236 - lr: 0.0010 - 337ms/epoch - 8ms/step Epoch 7/500 Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00007: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0564 - val_loss: 0.0531 - lr: 0.0010 - 356ms/epoch - 8ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0514 - val_loss: 0.0655 - lr: 1.0000e-04 - 344ms/epoch - 8ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0376 - val_loss: 0.0718 - lr: 1.0000e-04 - 330ms/epoch - 8ms/step Epoch 10/500 Epoch 00010: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0350 - val_loss: 0.0798 - lr: 1.0000e-04 - 407ms/epoch - 9ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0333 - val_loss: 0.0862 - lr: 1.0000e-04 - 420ms/epoch - 10ms/step Epoch 12/500 Epoch 00012: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00012: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0375 - val_loss: 0.0874 - lr: 1.0000e-04 - 372ms/epoch - 9ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0340 - val_loss: 0.0880 - lr: 1.0000e-05 - 438ms/epoch - 10ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0332 - val_loss: 0.0875 - lr: 1.0000e-05 - 370ms/epoch - 9ms/step Epoch 15/500 Epoch 00015: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0303 - val_loss: 0.0875 - lr: 1.0000e-05 - 359ms/epoch - 8ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0312 - val_loss: 0.0880 - lr: 1.0000e-05 - 430ms/epoch - 10ms/step Epoch 17/500 Epoch 00017: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00017: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0349 - val_loss: 0.0887 - lr: 1.0000e-05 - 362ms/epoch - 8ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0334 - val_loss: 0.0883 - lr: 1.0000e-05 - 363ms/epoch - 8ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0351 - val_loss: 0.0886 - lr: 1.0000e-05 - 341ms/epoch - 8ms/step Epoch 20/500 Epoch 00020: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0309 - val_loss: 0.0893 - lr: 1.0000e-05 - 376ms/epoch - 9ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0317 - val_loss: 0.0891 - lr: 1.0000e-05 - 464ms/epoch - 11ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0304 - val_loss: 0.0889 - lr: 1.0000e-05 - 402ms/epoch - 9ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0320 - val_loss: 0.0883 - lr: 1.0000e-05 - 392ms/epoch - 9ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0328 - val_loss: 0.0887 - lr: 1.0000e-05 - 392ms/epoch - 9ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0321 - val_loss: 0.0899 - lr: 1.0000e-05 - 349ms/epoch - 8ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0349 - val_loss: 0.0907 - lr: 1.0000e-05 - 355ms/epoch - 8ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0318 - val_loss: 0.0909 - lr: 1.0000e-05 - 377ms/epoch - 9ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0353 - val_loss: 0.0912 - lr: 1.0000e-05 - 334ms/epoch - 8ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0324 - val_loss: 0.0916 - lr: 1.0000e-05 - 402ms/epoch - 9ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0320 - val_loss: 0.0913 - lr: 1.0000e-05 - 362ms/epoch - 8ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0326 - val_loss: 0.0905 - lr: 1.0000e-05 - 387ms/epoch - 9ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0351 - val_loss: 0.0901 - lr: 1.0000e-05 - 393ms/epoch - 9ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0325 - val_loss: 0.0902 - lr: 1.0000e-05 - 375ms/epoch - 9ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0305 - val_loss: 0.0903 - lr: 1.0000e-05 - 360ms/epoch - 8ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0274 - val_loss: 0.0902 - lr: 1.0000e-05 - 420ms/epoch - 10ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0316 - val_loss: 0.0892 - lr: 1.0000e-05 - 463ms/epoch - 11ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0324 - val_loss: 0.0885 - lr: 1.0000e-05 - 431ms/epoch - 10ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0343 - val_loss: 0.0880 - lr: 1.0000e-05 - 374ms/epoch - 9ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0273 - val_loss: 0.0873 - lr: 1.0000e-05 - 362ms/epoch - 8ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0306 - val_loss: 0.0871 - lr: 1.0000e-05 - 368ms/epoch - 9ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0311 - val_loss: 0.0876 - lr: 1.0000e-05 - 349ms/epoch - 8ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0342 - val_loss: 0.0866 - lr: 1.0000e-05 - 422ms/epoch - 10ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0299 - val_loss: 0.0860 - lr: 1.0000e-05 - 397ms/epoch - 9ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0306 - val_loss: 0.0855 - lr: 1.0000e-05 - 425ms/epoch - 10ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0309 - val_loss: 0.0858 - lr: 1.0000e-05 - 336ms/epoch - 8ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0297 - val_loss: 0.0853 - lr: 1.0000e-05 - 347ms/epoch - 8ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0313 - val_loss: 0.0837 - lr: 1.0000e-05 - 383ms/epoch - 9ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0272 - val_loss: 0.0840 - lr: 1.0000e-05 - 411ms/epoch - 10ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0307 - val_loss: 0.0844 - lr: 1.0000e-05 - 440ms/epoch - 10ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0339 - val_loss: 0.0841 - lr: 1.0000e-05 - 355ms/epoch - 8ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0286 - val_loss: 0.0844 - lr: 1.0000e-05 - 431ms/epoch - 10ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.04661 43/43 - 0s - loss: 0.0311 - val_loss: 0.0838 - lr: 1.0000e-05 - 401ms/epoch - 9ms/step Epoch 00052: early stopping
SMA
Prediction vs Close: 51.87% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 39.29635655476263
RMSE: 6.268680607174258
MAPE: 5.07416987111494
EMA
Prediction vs Close: 56.34% Accuracy
Prediction vs Prediction: 48.88% Accuracy
MSE: 30.25057164247689
RMSE: 5.50005196725239
MAPE: 4.444486270049439
WMA
Prediction vs Close: 57.84% Accuracy
Prediction vs Prediction: 41.79% Accuracy
MSE: 64.92643731055051
RMSE: 8.057694292448089
MAPE: 6.289003841800032
DEMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 46.27% Accuracy
MSE: 52.98651621196421
RMSE: 7.279183760008
MAPE: 5.725540843661134
KAMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 33.6026987861483
RMSE: 5.796783486223054
MAPE: 4.518981487962124
MIDPOINT
Prediction vs Close: 50.0% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 22.7414116550439
RMSE: 4.768795618921396
MAPE: 3.944458615157319
T3
Prediction vs Close: 51.87% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 71.14601272964887
RMSE: 8.434809584670473
MAPE: 6.848574357624394
TEMA
TEMA([input_arrays], [timeperiod=30])
Triple Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
9
Working on TEMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.49 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4352.703, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3889.412, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.26 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3689.930, Time=0.06 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3574.245, Time=0.11 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.24 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.84 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3576.245, Time=0.22 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.299 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1783.123
Date: Sun, 12 Dec 2021 AIC 3574.245
Time: 13:32:51 BIC 3593.008
Sample: 0 HQIC 3581.451
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1480 0.004 -302.430 0.000 -1.155 -1.141
ar.L2 -0.8300 0.008 -99.682 0.000 -0.846 -0.814
ar.L3 -0.3687 0.007 -50.527 0.000 -0.383 -0.354
sigma2 4.9055 0.028 175.970 0.000 4.851 4.960
===================================================================================
Ljung-Box (L1) (Q): 11.61 Jarque-Bera (JB): 1261976.58
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.16 Skew: 2.52
Prob(H) (two-sided): 0.00 Kurtosis: 196.90
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
WARNING:tensorflow:Layer lstm_7 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
WARNING:tensorflow:Layer lstm_7 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.04929, saving model to LSTM1.h5 90/90 - 2s - loss: 0.2001 - val_loss: 0.0493 - lr: 0.0010 - 2s/epoch - 25ms/step Epoch 2/500 Epoch 00002: val_loss did not improve from 0.04929 90/90 - 1s - loss: 0.1000 - val_loss: 0.4574 - lr: 0.0010 - 742ms/epoch - 8ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.04929 90/90 - 1s - loss: 0.0796 - val_loss: 0.5206 - lr: 0.0010 - 774ms/epoch - 9ms/step Epoch 4/500 Epoch 00004: val_loss did not improve from 0.04929 90/90 - 1s - loss: 0.0593 - val_loss: 0.1143 - lr: 0.0010 - 742ms/epoch - 8ms/step Epoch 5/500 Epoch 00005: val_loss did not improve from 0.04929 90/90 - 1s - loss: 0.0503 - val_loss: 0.1298 - lr: 0.0010 - 730ms/epoch - 8ms/step Epoch 6/500 Epoch 00006: val_loss improved from 0.04929 to 0.02034, saving model to LSTM1.h5 90/90 - 1s - loss: 0.0385 - val_loss: 0.0203 - lr: 0.0010 - 902ms/epoch - 10ms/step Epoch 7/500 Epoch 00007: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0496 - val_loss: 0.3909 - lr: 0.0010 - 781ms/epoch - 9ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0404 - val_loss: 0.1832 - lr: 0.0010 - 752ms/epoch - 8ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0335 - val_loss: 0.0762 - lr: 0.0010 - 739ms/epoch - 8ms/step Epoch 10/500 Epoch 00010: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0276 - val_loss: 0.0831 - lr: 0.0010 - 758ms/epoch - 8ms/step Epoch 11/500 Epoch 00011: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00011: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0246 - val_loss: 0.0913 - lr: 0.0010 - 733ms/epoch - 8ms/step Epoch 12/500 Epoch 00012: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0346 - val_loss: 0.0851 - lr: 1.0000e-04 - 743ms/epoch - 8ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0245 - val_loss: 0.0867 - lr: 1.0000e-04 - 700ms/epoch - 8ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0208 - val_loss: 0.0913 - lr: 1.0000e-04 - 680ms/epoch - 8ms/step Epoch 15/500 Epoch 00015: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0193 - val_loss: 0.0880 - lr: 1.0000e-04 - 722ms/epoch - 8ms/step Epoch 16/500 Epoch 00016: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00016: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0218 - val_loss: 0.0926 - lr: 1.0000e-04 - 716ms/epoch - 8ms/step Epoch 17/500 Epoch 00017: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0230 - val_loss: 0.0940 - lr: 1.0000e-05 - 692ms/epoch - 8ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0201 - val_loss: 0.0942 - lr: 1.0000e-05 - 771ms/epoch - 9ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0188 - val_loss: 0.0936 - lr: 1.0000e-05 - 716ms/epoch - 8ms/step Epoch 20/500 Epoch 00020: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0195 - val_loss: 0.0932 - lr: 1.0000e-05 - 819ms/epoch - 9ms/step Epoch 21/500 Epoch 00021: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00021: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0195 - val_loss: 0.0932 - lr: 1.0000e-05 - 743ms/epoch - 8ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0219 - val_loss: 0.0940 - lr: 1.0000e-05 - 727ms/epoch - 8ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0211 - val_loss: 0.0941 - lr: 1.0000e-05 - 711ms/epoch - 8ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0209 - val_loss: 0.0942 - lr: 1.0000e-05 - 693ms/epoch - 8ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0205 - val_loss: 0.0958 - lr: 1.0000e-05 - 704ms/epoch - 8ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0199 - val_loss: 0.0967 - lr: 1.0000e-05 - 763ms/epoch - 8ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0206 - val_loss: 0.0963 - lr: 1.0000e-05 - 738ms/epoch - 8ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0205 - val_loss: 0.0947 - lr: 1.0000e-05 - 704ms/epoch - 8ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0201 - val_loss: 0.0938 - lr: 1.0000e-05 - 725ms/epoch - 8ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0212 - val_loss: 0.0931 - lr: 1.0000e-05 - 728ms/epoch - 8ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0192 - val_loss: 0.0921 - lr: 1.0000e-05 - 726ms/epoch - 8ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0213 - val_loss: 0.0924 - lr: 1.0000e-05 - 685ms/epoch - 8ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0199 - val_loss: 0.0932 - lr: 1.0000e-05 - 704ms/epoch - 8ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0195 - val_loss: 0.0930 - lr: 1.0000e-05 - 774ms/epoch - 9ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0188 - val_loss: 0.0930 - lr: 1.0000e-05 - 712ms/epoch - 8ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0194 - val_loss: 0.0922 - lr: 1.0000e-05 - 697ms/epoch - 8ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0182 - val_loss: 0.0928 - lr: 1.0000e-05 - 699ms/epoch - 8ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0199 - val_loss: 0.0923 - lr: 1.0000e-05 - 708ms/epoch - 8ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0178 - val_loss: 0.0904 - lr: 1.0000e-05 - 729ms/epoch - 8ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0201 - val_loss: 0.0898 - lr: 1.0000e-05 - 729ms/epoch - 8ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0203 - val_loss: 0.0895 - lr: 1.0000e-05 - 734ms/epoch - 8ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0198 - val_loss: 0.0913 - lr: 1.0000e-05 - 765ms/epoch - 8ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0202 - val_loss: 0.0894 - lr: 1.0000e-05 - 722ms/epoch - 8ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0198 - val_loss: 0.0917 - lr: 1.0000e-05 - 733ms/epoch - 8ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0215 - val_loss: 0.0926 - lr: 1.0000e-05 - 730ms/epoch - 8ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0197 - val_loss: 0.0949 - lr: 1.0000e-05 - 786ms/epoch - 9ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0193 - val_loss: 0.0944 - lr: 1.0000e-05 - 749ms/epoch - 8ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0171 - val_loss: 0.0946 - lr: 1.0000e-05 - 710ms/epoch - 8ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0193 - val_loss: 0.0973 - lr: 1.0000e-05 - 788ms/epoch - 9ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0196 - val_loss: 0.0998 - lr: 1.0000e-05 - 741ms/epoch - 8ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0199 - val_loss: 0.1008 - lr: 1.0000e-05 - 741ms/epoch - 8ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0202 - val_loss: 0.1000 - lr: 1.0000e-05 - 730ms/epoch - 8ms/step Epoch 53/500 Epoch 00053: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0192 - val_loss: 0.1023 - lr: 1.0000e-05 - 725ms/epoch - 8ms/step Epoch 54/500 Epoch 00054: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0202 - val_loss: 0.1038 - lr: 1.0000e-05 - 813ms/epoch - 9ms/step Epoch 55/500 Epoch 00055: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0170 - val_loss: 0.1030 - lr: 1.0000e-05 - 725ms/epoch - 8ms/step Epoch 56/500 Epoch 00056: val_loss did not improve from 0.02034 90/90 - 1s - loss: 0.0201 - val_loss: 0.1042 - lr: 1.0000e-05 - 788ms/epoch - 9ms/step Epoch 00056: early stopping
SMA Prediction vs Close: 51.87% Accuracy Prediction vs Prediction: 45.9% Accuracy MSE: 39.29635655476263 RMSE: 6.268680607174258 MAPE: 5.07416987111494 EMA Prediction vs Close: 56.34% Accuracy Prediction vs Prediction: 48.88% Accuracy MSE: 30.25057164247689 RMSE: 5.50005196725239 MAPE: 4.444486270049439 WMA Prediction vs Close: 57.84% Accuracy Prediction vs Prediction: 41.79% Accuracy MSE: 64.92643731055051 RMSE: 8.057694292448089 MAPE: 6.289003841800032 DEMA Prediction vs Close: 52.61% Accuracy Prediction vs Prediction: 46.27% Accuracy MSE: 52.98651621196421 RMSE: 7.279183760008 MAPE: 5.725540843661134 KAMA Prediction vs Close: 54.48% Accuracy Prediction vs Prediction: 48.51% Accuracy MSE: 33.6026987861483 RMSE: 5.796783486223054 MAPE: 4.518981487962124 MIDPOINT Prediction vs Close: 50.0% Accuracy Prediction vs Prediction: 46.64% Accuracy MSE: 22.7414116550439 RMSE: 4.768795618921396 MAPE: 3.944458615157319 T3 Prediction vs Close: 51.87% Accuracy Prediction vs Prediction: 46.64% Accuracy MSE: 71.14601272964887 RMSE: 8.434809584670473 MAPE: 6.848574357624394 TEMA Prediction vs Close: 51.12% Accuracy Prediction vs Prediction: 49.25% Accuracy MSE: 26.608302345887367 RMSE: 5.158323598407468 MAPE: 4.336839144940602 Runtime: mins: 13.275118494316668
Architecture used¶
In [85]:
from google.colab import files
import cv2
uploaded = files.upload()
Saving Experiment1.png to Experiment1 (1).png
In [86]:
img = cv2.imread('Experiment1.png')
plt.figure(figsize=(20,10))
plt.axis("off")
plt.title('LSTM Architecture '+imgfile,fontsize=18)
plt.imshow(img)
Out[86]:
<matplotlib.image.AxesImage at 0x7f75c3c67b10>
Excess kurtosis is a metric that compares the kurtosis of a distribution against the kurtosis of a normal distribution. The kurtosis of a normal distribution equals 3. Therefore, the excess kurtosis is found using the formula below:
Excess Kurtosis = Kurtosis – 3
Model Plots¶
In [87]:
np.save("X_train_appl.npy", X_train)
np.save("y_train_appl.npy", y_train)
np.save("X_test_appl.npy", X_test)
np.save("y_test_appl.npy", y_test)
np.save("yc_train_appl.npy", yc_train)
np.save("yc_test_appl.npy", yc_test)
np.save('index_train_appl.npy', index_train)
np.save('index_test_appl.npy', index_test)
In [88]:
list(simulation1.keys())
Out[88]:
['SMA', 'EMA', 'WMA', 'DEMA', 'KAMA', 'MIDPOINT', 'T3', 'TEMA']
In [97]:
import json
with open('simulation1_data.json') as json_file:
simulation1 = json.load(json_file)
fileimg = 'Experiment1'
In [98]:
for i in range(len(list(simulation1.keys()))):
SIM = list(simulation1.keys())[i]
plot_train(simulation1,SIM)
plot_test(simulation1,SIM)
----- Train RMSE for SMA ----- 8.0413849019624 ----- Train_MSE_LSTM for SMA ----- 64.66387114150885 ----- Train MAE LSTM for SMA ----- 6.984172090422362
----- Test RMSE for SMA----- 6.268680607174258 ----- Test_MSE_LSTM for SMA----- 39.29635655476263 ----- Test_MAE_LSTM for SMA----- 5.07416987111494
----- Train RMSE for EMA ----- 9.093795784259413 ----- Train_MSE_LSTM for EMA ----- 82.69712176581427 ----- Train MAE LSTM for EMA ----- 7.9494118779796255
----- Test RMSE for EMA----- 5.50005196725239 ----- Test_MSE_LSTM for EMA----- 30.25057164247689 ----- Test_MAE_LSTM for EMA----- 4.444486270049439
----- Train RMSE for WMA ----- 9.540832646041645 ----- Train_MSE_LSTM for WMA ----- 91.02748757977402 ----- Train MAE LSTM for WMA ----- 8.461359613734103
----- Test RMSE for WMA----- 8.057694292448089 ----- Test_MSE_LSTM for WMA----- 64.92643731055051 ----- Test_MAE_LSTM for WMA----- 6.289003841800032
----- Train RMSE for DEMA ----- 11.100597362863033 ----- Train_MSE_LSTM for DEMA ----- 123.2232618124017 ----- Train MAE LSTM for DEMA ----- 9.911796547735277
----- Test RMSE for DEMA----- 7.279183760008 ----- Test_MSE_LSTM for DEMA----- 52.98651621196421 ----- Test_MAE_LSTM for DEMA----- 5.725540843661134
----- Train RMSE for KAMA ----- 9.711048663776115 ----- Train_MSE_LSTM for KAMA ----- 94.30446615022788 ----- Train MAE LSTM for KAMA ----- 8.63911047031167
----- Test RMSE for KAMA----- 5.796783486223054 ----- Test_MSE_LSTM for KAMA----- 33.6026987861483 ----- Test_MAE_LSTM for KAMA----- 4.518981487962124
----- Train RMSE for MIDPOINT ----- 8.42723685446347 ----- Train_MSE_LSTM for MIDPOINT ----- 71.01832100122736 ----- Train MAE LSTM for MIDPOINT ----- 7.433581155738655
----- Test RMSE for MIDPOINT----- 4.768795618921396 ----- Test_MSE_LSTM for MIDPOINT----- 22.7414116550439 ----- Test_MAE_LSTM for MIDPOINT----- 3.944458615157319
----- Train RMSE for T3 ----- 10.78572680059317 ----- Train_MSE_LSTM for T3 ----- 116.33190261703379 ----- Train MAE LSTM for T3 ----- 9.683946313650006
----- Test RMSE for T3----- 8.434809584670473 ----- Test_MSE_LSTM for T3----- 71.14601272964887 ----- Test_MAE_LSTM for T3----- 6.848574357624394
----- Train RMSE for TEMA ----- 6.938756727032321 ----- Train_MSE_LSTM for TEMA ----- 48.14634491693629 ----- Train MAE LSTM for TEMA ----- 4.664446043887978
----- Test RMSE for TEMA----- 5.158323598407468 ----- Test_MSE_LSTM for TEMA----- 26.608302345887367 ----- Test_MAE_LSTM for TEMA----- 4.336839144940602
Univariate Arima Multistep MutiVariate LSTM Hybrid Model Experiment 2¶
In [96]:
def get_lstm(data,original_data, train_len, test_len,img_file,ma ,lstm_len=3):
# prepare train and test data
X_value = pd.DataFrame(data.iloc[:, :])
y_value = pd.DataFrame(data.iloc[:, 3])
X_scaler = MinMaxScaler(feature_range=(-1, 1))
y_scaler = MinMaxScaler(feature_range=(-1, 1))
X_scaler.fit(X_value)
y_scaler.fit(y_value)
# Get data and check shape
X, y, yc = get_X_y(X_scale_dataset, y_scale_dataset)#X will be of shape 224 X 3 X 21 (each 3 X 21 array will be 3 days' worth of data). yc will have the corresponding closing price value
# pdb.set_trace()
X_train, X_test, = split_train_test(X)
y_train, y_test, = split_train_test(y)
# yc_train, yc_test, = split_train_test(original_data)
index_train, index_test, = predict_index(dataset_final, X_train, n_steps_in, n_steps_out)
det = 20
input_dim = X_train.shape[1]#3
feature_size = X_train.shape[2]#24
output_dim = y_train.shape[1]#1
# # Option 1
# # Set up & fit LSTM RNN
# model = Sequential()
# model.add(LSTM(256, activation='relu', kernel_initializer='he_normal', input_shape=(input_dim, feature_size)))
# model.add(Dense(units=64,activation='relu'))
# model.add(Dropout(0.5))
# model.add(Dense(units=output_dim))
# model.compile(optimizer=Adam(learning_rate = 0.001), loss='mse')
# ## Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# option 2
model = Sequential()
model.add(Bidirectional(LSTM(units= 128), input_shape=(input_dim, feature_size)))
model.add(Dense(64))
model.add(Dense(units=output_dim))
model.compile(optimizer=Adam(learning_rate = 0.001), loss='mean_squared_error', metrics=['accuracy'])
# Common code
callbacks = [
EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
ModelCheckpoint('LSTM2.h5', verbose=1, save_best_only=True, save_weights_only=True)]
fname1 = img_file+'.png'
tensorflow.keras.utils.plot_model(
model, to_file=fname1, show_shapes=True, show_dtype=False,
show_layer_names=True, expand_nested=False, dpi=96,
layer_range=None, show_layer_activations=False
)
history = model.fit(X_train, y_train, epochs=500, batch_size=int( optimized_period[ma]), verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# plot loss
fname2 = img_file+'-'+ma
plt.title(img_file+'-'+ma+' Loss')
plt.xlabel("Epochs")
plt.ylabel("Loss")
pyplot.plot(history.history['loss'], label='train')
pyplot.plot(history.history['val_loss'], label='validation')
pyplot.legend()
pyplot.savefig(fname2+'.png',dpi='figure')
pyplot.show()
# Option 3
# define custom activation
#
# class Double_Tanh(Activation):
# def __init__(self, activation, **kwargs):
# super(Double_Tanh, self).__init__(activation, **kwargs)
# self.__name__ = 'double_tanh'
# def double_tanh(x):
# return (K.tanh(x) * 2)
# get_custom_objects().update({'double_tanh':Double_Tanh(double_tanh)})
# # Model Generation
# model = Sequential()
# #check https://machinelearningmastery.com/use-weight-regularization-lstm-networks-time-series-forecasting/
# model.add(LSTM(25, input_shape=(input_dim, feature_size), dropout=0.2, kernel_regularizer=l1_l2(0.00,0.00), bias_regularizer=l1_l2(0.00,0.00)))
# model.add(Dense(1))
# model.add(Activation(double_tanh))
# model.compile(loss='mean_squared_error', optimizer='adam', metrics=['mse', 'mae'])
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# Option 4
# Set up & fit LSTM RNN
# model = Sequential()
# model.add(LSTM(units=lstm_len, return_sequences=True, input_shape=(x_train.shape[1], 1)))
# model.add(LSTM(units=int(lstm_len/2)))
# model.add(Dense(1, activation='sigmoid'))
# model.compile(loss='mean_squared_error', optimizer='adam')
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# Generate predictions
predictiontr = model.predict(X_train, verbose=0)
predictiontr = y_scaler.inverse_transform(predictiontr).tolist()
outputtr = []
for i in range(len(predictiontr)):
outputtr.extend(predictiontr[i])
predictiontr = outputtr
# Generate error data
## replace with yc , xtest generated by new multistep method
mse_tr = mean_squared_error(y_train, predictiontr)
rmse_tr = mse_tr ** 0.5
# mape = mean_absolute_percentage_error(X_test, pd.Series(predictiontr))
mae_tr = mean_absolute_error(y_train, pd.Series(predictiontr))
# Original_tr = pd.Series(yc_train)
Original_tr = y_scaler.inverse_transform(y_train).flatten().tolist()
predictionte = model.predict(X_test, verbose=0)
predictionte = (y_scaler.inverse_transform(predictionte)-det).tolist()
outputte = []
for i in range(len(predictionte)):
outputte.extend(predictionte[i])
predictionte = outputte
# Generate error data
mse_te = mean_squared_error(y_test, predictionte)
rmse_te = mse_te ** 0.5
# mape = mean_absolute_percentage_error(X_test, pd.Series(predictionte))
mae_te = mean_absolute_error(y_test, pd.Series(predictionte))
# Original_te = pd.Series(yc_test)
Original_te = y_scaler.inverse_transform(y_test).flatten().tolist()
return Original_tr, predictiontr, mse_tr, rmse_tr,mae_tr,Original_te,predictionte, mse_te,rmse_te,mae_te
In [97]:
if __name__ == '__main__':
start_time = timeit.default_timer()
simulation2 = {}
imgfile = 'Experiment2'
for ma in optimized_period:
print(ma)
print(functions[ma])
print ( int( optimized_period[ma]))
# if ma == 'SMA':
low_vol = df.apply(lambda c: functions[ma](c, timeperiod = int( optimized_period[ma])))
low_vol = low_vol.fillna(0)
low_vol_data = df['close']
high_vol = pd.DataFrame()
df2 = df.copy()
for i in df2.columns:
if i in low_vol.columns:
high_vol[i] = df2[i].subtract(low_vol[i], fill_value=0)
high_vol_data = df['close']
## *****************************************************
# Generate ARIMA and LSTM predictions
print('\nWorking on ' + ma + ' predictions')
try:
print('parameters used : ', train_len, test_len)
low_vol_Original, low_vol_prediction, low_vol_mse, low_vol_rmse,low_vol_mae = get_arima(low_vol,low_vol_data, train_len, test_len)
except:
print('ARIMA error, skipping to next MA type')
continue
Original_tr, predictiontr, mse_tr, rmse_tr,mae_tr,high_vol_Original, high_vol_prediction, high_vol_mse, high_vol_rmse,high_vol_mae, = get_lstm(high_vol,high_vol_data, train_len, test_len,imgfile,ma)
final_prediction_tr = df['close'].head(train_len).values + pd.Series(predictiontr) # ignoring first 3 steps
mse_ftr = mean_squared_error(df['close'].head(train_len).values,final_prediction_tr.values)
rmse_ftr = mse_ftr ** 0.5
mape_ftr = mean_absolute_percentage_error(df['close'].head(train_len).reset_index(drop=True), final_prediction_tr)
mae_ftr = mean_absolute_error(df['close'].head(train_len).reset_index(drop=True), final_prediction_tr)
final_prediction = pd.Series(low_vol_prediction[3:]) + pd.Series(high_vol_prediction)
mse = mean_squared_error(df['close'].tail(test_len).values,final_prediction.values)
rmse = mse ** 0.5
mape = mean_absolute_percentage_error(df['close'].tail(test_len).reset_index(drop=True), final_prediction)
mae = mean_absolute_error(df['close'].tail(test_len).reset_index(drop=True), final_prediction)
# Generate prediction accuracy
actual = df['close'].tail(test_len).values
result_1 = []
result_2 = []
for i in range(1, len(final_prediction)):
# Compare prediction to previous close price
if final_prediction[i] > actual[i-1] and actual[i] > actual[i-1]:
result_1.append(1)
elif final_prediction[i] < actual[i-1] and actual[i] < actual[i-1]:
result_1.append(1)
else:
result_1.append(0)
# Compare prediction to previous prediction
if final_prediction[i] > final_prediction[i-1] and actual[i] > actual[i-1]:
result_2.append(1)
elif final_prediction[i] < final_prediction[i-1] and actual[i] < actual[i-1]:
result_2.append(1)
else:
result_2.append(0)
accuracy_1 = np.mean(result_1)
accuracy_2 = np.mean(result_2)
simulation2[ma] = {'low_vol': {'original':list(low_vol_Original), 'prediction': list(low_vol_prediction) , 'mse': low_vol_mse,
'rmse': low_vol_rmse, 'mae' : low_vol_mae},
'high_vol': {'original':list(high_vol_Original),'prediction': list(high_vol_prediction), 'mse': high_vol_mse,
'rmse': high_vol_rmse, 'mae' : high_vol_mae},
'final_tr': {'original':df['close'].head(train_len).tolist(),'prediction': final_prediction_tr.values.tolist(), 'mse': mse_ftr,
'rmse': rmse_ftr, 'mae' : mae_ftr},
'final': {'original': df['close'].tail(test_len).tolist(), 'prediction': final_prediction.values.tolist(), 'mse': mse,
'rmse': rmse, 'mae': mae },
'accuracy': {'prediction vs close': accuracy_1, 'prediction vs prediction': accuracy_2}}
# save simulation data here as checkpoint
with open('simulation2_data.json', 'w') as fp:
json.dump(simulation2, fp)
for ma in simulation2.keys():
print('\n' + ma)
print('Prediction vs Close:\t\t' + str(round(100*simulation2[ma]['accuracy']['prediction vs close'], 2))
+ '% Accuracy')
print('Prediction vs Prediction:\t' + str(round(100*simulation2[ma]['accuracy']['prediction vs prediction'], 2))
+ '% Accuracy')
print('MSE:\t', simulation2[ma]['final']['mse'],
'\nRMSE:\t', simulation2[ma]['final']['rmse'],
'\nMAPE:\t', simulation2[ma]['final']['mae'])#,
# '\nMAPE:\t', simulation[ma]['final']['mape'])
# else:
# break
elapsed = timeit.default_timer() - start_time
print('Runtime: mins:',elapsed/60)
SMA
SMA([input_arrays], [timeperiod=30])
Simple Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
17
Working on SMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.49 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4157.020, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3687.148, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.21 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3458.651, Time=0.08 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3322.133, Time=0.08 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=0.76 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.86 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3324.133, Time=0.24 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 2.790 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1657.067
Date: Sun, 12 Dec 2021 AIC 3322.133
Time: 14:00:25 BIC 3340.897
Sample: 0 HQIC 3329.339
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1966 0.003 -387.226 0.000 -1.203 -1.191
ar.L2 -0.8952 0.006 -138.692 0.000 -0.908 -0.883
ar.L3 -0.3968 0.006 -68.284 0.000 -0.408 -0.385
sigma2 3.5858 0.017 214.535 0.000 3.553 3.619
===================================================================================
Ljung-Box (L1) (Q): 14.47 Jarque-Bera (JB): 2428881.42
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.90
Prob(H) (two-sided): 0.00 Kurtosis: 271.99
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.06118, saving model to LSTM2.h5
48/48 - 4s - loss: 0.1411 - accuracy: 0.0000e+00 - val_loss: 0.0612 - val_accuracy: 0.0037 - lr: 0.0010 - 4s/epoch - 82ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.06118
48/48 - 0s - loss: 0.0648 - accuracy: 0.0000e+00 - val_loss: 0.0751 - val_accuracy: 0.0037 - lr: 0.0010 - 259ms/epoch - 5ms/step
Epoch 3/500
Epoch 00003: val_loss improved from 0.06118 to 0.01076, saving model to LSTM2.h5
48/48 - 0s - loss: 0.0226 - accuracy: 0.0000e+00 - val_loss: 0.0108 - val_accuracy: 0.0037 - lr: 0.0010 - 300ms/epoch - 6ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.01076
48/48 - 0s - loss: 0.0058 - accuracy: 0.0000e+00 - val_loss: 0.0202 - val_accuracy: 0.0037 - lr: 0.0010 - 280ms/epoch - 6ms/step
Epoch 5/500
Epoch 00005: val_loss improved from 0.01076 to 0.00547, saving model to LSTM2.h5
48/48 - 0s - loss: 0.0055 - accuracy: 0.0000e+00 - val_loss: 0.0055 - val_accuracy: 0.0037 - lr: 0.0010 - 324ms/epoch - 7ms/step
Epoch 6/500
Epoch 00006: val_loss did not improve from 0.00547
48/48 - 0s - loss: 0.0022 - accuracy: 0.0000e+00 - val_loss: 0.0057 - val_accuracy: 0.0037 - lr: 0.0010 - 280ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: val_loss improved from 0.00547 to 0.00540, saving model to LSTM2.h5
48/48 - 0s - loss: 0.0016 - accuracy: 0.0000e+00 - val_loss: 0.0054 - val_accuracy: 0.0037 - lr: 0.0010 - 289ms/epoch - 6ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.00540
48/48 - 0s - loss: 0.0015 - accuracy: 0.0000e+00 - val_loss: 0.0062 - val_accuracy: 0.0037 - lr: 0.0010 - 270ms/epoch - 6ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.00540
48/48 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0100 - val_accuracy: 0.0037 - lr: 0.0010 - 264ms/epoch - 6ms/step
Epoch 10/500
Epoch 00010: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00010: val_loss did not improve from 0.00540
48/48 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0128 - val_accuracy: 0.0037 - lr: 0.0010 - 276ms/epoch - 6ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.00540
48/48 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 262ms/epoch - 5ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.00540
48/48 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0158 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 288ms/epoch - 6ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.00540
48/48 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0170 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 286ms/epoch - 6ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.00540
48/48 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0182 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 292ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00015: val_loss did not improve from 0.00540
48/48 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0193 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 286ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.00540
48/48 - 0s - loss: 9.1783e-04 - accuracy: 0.0000e+00 - val_loss: 0.0193 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 283ms/epoch - 6ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.00540
48/48 - 0s - loss: 9.1492e-04 - accuracy: 0.0000e+00 - val_loss: 0.0193 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 236ms/epoch - 5ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.00540
48/48 - 0s - loss: 9.1247e-04 - accuracy: 0.0000e+00 - val_loss: 0.0193 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 269ms/epoch - 6ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.00540
48/48 - 0s - loss: 9.1016e-04 - accuracy: 0.0000e+00 - val_loss: 0.0194 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 271ms/epoch - 6ms/step
Epoch 20/500
Epoch 00020: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00020: val_loss did not improve from 0.00540
48/48 - 0s - loss: 9.0794e-04 - accuracy: 0.0000e+00 - val_loss: 0.0194 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 249ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.00540
48/48 - 0s - loss: 9.0577e-04 - accuracy: 0.0000e+00 - val_loss: 0.0195 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 284ms/epoch - 6ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.00540
48/48 - 0s - loss: 9.0365e-04 - accuracy: 0.0000e+00 - val_loss: 0.0195 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 294ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.00540
48/48 - 0s - loss: 9.0156e-04 - accuracy: 0.0000e+00 - val_loss: 0.0196 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 293ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.9952e-04 - accuracy: 0.0000e+00 - val_loss: 0.0197 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 252ms/epoch - 5ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.9750e-04 - accuracy: 0.0000e+00 - val_loss: 0.0198 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 280ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.9550e-04 - accuracy: 0.0000e+00 - val_loss: 0.0198 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 279ms/epoch - 6ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.9353e-04 - accuracy: 0.0000e+00 - val_loss: 0.0199 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 275ms/epoch - 6ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.9158e-04 - accuracy: 0.0000e+00 - val_loss: 0.0200 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 274ms/epoch - 6ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.8965e-04 - accuracy: 0.0000e+00 - val_loss: 0.0201 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 295ms/epoch - 6ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.8772e-04 - accuracy: 0.0000e+00 - val_loss: 0.0202 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 319ms/epoch - 7ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.8581e-04 - accuracy: 0.0000e+00 - val_loss: 0.0203 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 250ms/epoch - 5ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.8390e-04 - accuracy: 0.0000e+00 - val_loss: 0.0204 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 279ms/epoch - 6ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.8200e-04 - accuracy: 0.0000e+00 - val_loss: 0.0205 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 239ms/epoch - 5ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.8011e-04 - accuracy: 0.0000e+00 - val_loss: 0.0206 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 286ms/epoch - 6ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.7822e-04 - accuracy: 0.0000e+00 - val_loss: 0.0207 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 242ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.7632e-04 - accuracy: 0.0000e+00 - val_loss: 0.0209 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 254ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.7443e-04 - accuracy: 0.0000e+00 - val_loss: 0.0210 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 275ms/epoch - 6ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.7254e-04 - accuracy: 0.0000e+00 - val_loss: 0.0211 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 284ms/epoch - 6ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.7065e-04 - accuracy: 0.0000e+00 - val_loss: 0.0212 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 269ms/epoch - 6ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.6876e-04 - accuracy: 0.0000e+00 - val_loss: 0.0213 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 280ms/epoch - 6ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.6687e-04 - accuracy: 0.0000e+00 - val_loss: 0.0215 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 298ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.6497e-04 - accuracy: 0.0000e+00 - val_loss: 0.0216 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 251ms/epoch - 5ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.6308e-04 - accuracy: 0.0000e+00 - val_loss: 0.0217 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 321ms/epoch - 7ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.6118e-04 - accuracy: 0.0000e+00 - val_loss: 0.0219 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 262ms/epoch - 5ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.5929e-04 - accuracy: 0.0000e+00 - val_loss: 0.0220 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 256ms/epoch - 5ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.5739e-04 - accuracy: 0.0000e+00 - val_loss: 0.0221 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 271ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.5550e-04 - accuracy: 0.0000e+00 - val_loss: 0.0223 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 310ms/epoch - 6ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.5360e-04 - accuracy: 0.0000e+00 - val_loss: 0.0224 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 265ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.5170e-04 - accuracy: 0.0000e+00 - val_loss: 0.0225 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 293ms/epoch - 6ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.4981e-04 - accuracy: 0.0000e+00 - val_loss: 0.0227 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 286ms/epoch - 6ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.4791e-04 - accuracy: 0.0000e+00 - val_loss: 0.0228 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 307ms/epoch - 6ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.4602e-04 - accuracy: 0.0000e+00 - val_loss: 0.0229 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 260ms/epoch - 5ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.4413e-04 - accuracy: 0.0000e+00 - val_loss: 0.0231 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 265ms/epoch - 6ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.4224e-04 - accuracy: 0.0000e+00 - val_loss: 0.0232 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 291ms/epoch - 6ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.4036e-04 - accuracy: 0.0000e+00 - val_loss: 0.0233 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 277ms/epoch - 6ms/step
Epoch 56/500
Epoch 00056: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.3847e-04 - accuracy: 0.0000e+00 - val_loss: 0.0235 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 277ms/epoch - 6ms/step
Epoch 57/500
Epoch 00057: val_loss did not improve from 0.00540
48/48 - 0s - loss: 8.3659e-04 - accuracy: 0.0000e+00 - val_loss: 0.0236 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 274ms/epoch - 6ms/step
Epoch 00057: early stopping
SMA
Prediction vs Close: 53.73% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 39.87524298067742
RMSE: 6.3146847095225125
MAPE: 5.088204561858354
EMA
EMA([input_arrays], [timeperiod=30])
Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
51
Working on EMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.41 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4231.556, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3761.238, Time=0.05 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.28 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3532.227, Time=0.08 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3394.496, Time=0.10 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=0.81 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.66 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3396.496, Time=0.22 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 2.636 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1693.248
Date: Sun, 12 Dec 2021 AIC 3394.496
Time: 14:01:56 BIC 3413.260
Sample: 0 HQIC 3401.702
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1982 0.003 -389.569 0.000 -1.204 -1.192
ar.L2 -0.8976 0.006 -139.811 0.000 -0.910 -0.885
ar.L3 -0.3984 0.006 -68.662 0.000 -0.410 -0.387
sigma2 3.9230 0.018 215.372 0.000 3.887 3.959
===================================================================================
Ljung-Box (L1) (Q): 14.54 Jarque-Bera (JB): 2462173.05
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.90
Prob(H) (two-sided): 0.00 Kurtosis: 273.82
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.04789, saving model to LSTM2.h5
16/16 - 4s - loss: 0.1362 - accuracy: 0.0000e+00 - val_loss: 0.0479 - val_accuracy: 0.0037 - lr: 0.0010 - 4s/epoch - 234ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.04789 to 0.01125, saving model to LSTM2.h5
16/16 - 0s - loss: 0.0586 - accuracy: 0.0000e+00 - val_loss: 0.0112 - val_accuracy: 0.0037 - lr: 0.0010 - 126ms/epoch - 8ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.01125
16/16 - 0s - loss: 0.0240 - accuracy: 0.0000e+00 - val_loss: 0.0633 - val_accuracy: 0.0037 - lr: 0.0010 - 111ms/epoch - 7ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.01125
16/16 - 0s - loss: 0.0143 - accuracy: 0.0000e+00 - val_loss: 0.0290 - val_accuracy: 0.0037 - lr: 0.0010 - 100ms/epoch - 6ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.01125
16/16 - 0s - loss: 0.0108 - accuracy: 0.0000e+00 - val_loss: 0.0113 - val_accuracy: 0.0037 - lr: 0.0010 - 102ms/epoch - 6ms/step
Epoch 6/500
Epoch 00006: val_loss improved from 0.01125 to 0.00912, saving model to LSTM2.h5
16/16 - 0s - loss: 0.0236 - accuracy: 0.0000e+00 - val_loss: 0.0091 - val_accuracy: 0.0037 - lr: 0.0010 - 124ms/epoch - 8ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.00912
16/16 - 0s - loss: 0.0040 - accuracy: 0.0000e+00 - val_loss: 0.0363 - val_accuracy: 0.0037 - lr: 0.0010 - 96ms/epoch - 6ms/step
Epoch 8/500
Epoch 00008: val_loss improved from 0.00912 to 0.00408, saving model to LSTM2.h5
16/16 - 0s - loss: 0.0155 - accuracy: 0.0000e+00 - val_loss: 0.0041 - val_accuracy: 0.0037 - lr: 0.0010 - 128ms/epoch - 8ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0078 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 0.0010 - 125ms/epoch - 8ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0150 - accuracy: 0.0000e+00 - val_loss: 0.0163 - val_accuracy: 0.0037 - lr: 0.0010 - 119ms/epoch - 7ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0069 - accuracy: 0.0000e+00 - val_loss: 0.0181 - val_accuracy: 0.0037 - lr: 0.0010 - 117ms/epoch - 7ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0113 - accuracy: 0.0000e+00 - val_loss: 0.0098 - val_accuracy: 0.0037 - lr: 0.0010 - 107ms/epoch - 7ms/step
Epoch 13/500
Epoch 00013: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00013: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0124 - accuracy: 0.0000e+00 - val_loss: 0.0073 - val_accuracy: 0.0037 - lr: 0.0010 - 113ms/epoch - 7ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0104 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 98ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0017 - accuracy: 0.0000e+00 - val_loss: 0.0049 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 109ms/epoch - 7ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 115ms/epoch - 7ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 99ms/epoch - 6ms/step
Epoch 18/500
Epoch 00018: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00018: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 135ms/epoch - 8ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 114ms/epoch - 7ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 108ms/epoch - 7ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 98ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00023: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 101ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 104ms/epoch - 7ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 119ms/epoch - 7ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 106ms/epoch - 7ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 118ms/epoch - 7ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 107ms/epoch - 7ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 112ms/epoch - 7ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 97ms/epoch - 6ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.00408
16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 98ms/epoch - 6ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.9942e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 101ms/epoch - 6ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.9609e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 130ms/epoch - 8ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.9281e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 108ms/epoch - 7ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.8958e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 109ms/epoch - 7ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.8640e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 98ms/epoch - 6ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.8326e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.8018e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 98ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.7714e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.7416e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.7123e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 108ms/epoch - 7ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.6835e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 105ms/epoch - 7ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.6552e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 115ms/epoch - 7ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.6274e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 109ms/epoch - 7ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.6001e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 112ms/epoch - 7ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.5734e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.5472e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 105ms/epoch - 7ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.5214e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.4962e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 113ms/epoch - 7ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.4715e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 100ms/epoch - 6ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.4472e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 113ms/epoch - 7ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.4235e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step
Epoch 56/500
Epoch 00056: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.4002e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 131ms/epoch - 8ms/step
Epoch 57/500
Epoch 00057: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.3773e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 114ms/epoch - 7ms/step
Epoch 58/500
Epoch 00058: val_loss did not improve from 0.00408
16/16 - 0s - loss: 9.3550e-04 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 00058: early stopping
SMA
Prediction vs Close: 53.73% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 39.87524298067742
RMSE: 6.3146847095225125
MAPE: 5.088204561858354
EMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 43.66% Accuracy
MSE: 73.6938994303845
RMSE: 8.584515095821342
MAPE: 7.207683534137507
WMA
WMA([input_arrays], [timeperiod=30])
Weighted Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
49
Working on WMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.40 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4264.089, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3793.930, Time=0.05 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.25 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3564.923, Time=0.08 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3427.258, Time=0.09 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.26 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.48 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3429.258, Time=0.19 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 2.840 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1709.629
Date: Sun, 12 Dec 2021 AIC 3427.258
Time: 14:03:16 BIC 3446.021
Sample: 0 HQIC 3434.464
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1981 0.003 -389.386 0.000 -1.204 -1.192
ar.L2 -0.8974 0.006 -139.699 0.000 -0.910 -0.885
ar.L3 -0.3983 0.006 -68.737 0.000 -0.410 -0.387
sigma2 4.0860 0.019 215.311 0.000 4.049 4.123
===================================================================================
Ljung-Box (L1) (Q): 14.57 Jarque-Bera (JB): 2460901.70
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.90
Prob(H) (two-sided): 0.00 Kurtosis: 273.75
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.00916, saving model to LSTM2.h5
17/17 - 4s - loss: 0.0891 - accuracy: 0.0000e+00 - val_loss: 0.0092 - val_accuracy: 0.0037 - lr: 0.0010 - 4s/epoch - 237ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.00916
17/17 - 0s - loss: 0.0709 - accuracy: 0.0000e+00 - val_loss: 0.0856 - val_accuracy: 0.0000e+00 - lr: 0.0010 - 102ms/epoch - 6ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.00916
17/17 - 0s - loss: 0.0175 - accuracy: 0.0000e+00 - val_loss: 0.0689 - val_accuracy: 0.0037 - lr: 0.0010 - 114ms/epoch - 7ms/step
Epoch 4/500
Epoch 00004: val_loss improved from 0.00916 to 0.00554, saving model to LSTM2.h5
17/17 - 0s - loss: 0.0238 - accuracy: 0.0000e+00 - val_loss: 0.0055 - val_accuracy: 0.0037 - lr: 0.0010 - 143ms/epoch - 8ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.00554
17/17 - 0s - loss: 0.0163 - accuracy: 0.0000e+00 - val_loss: 0.0060 - val_accuracy: 0.0037 - lr: 0.0010 - 122ms/epoch - 7ms/step
Epoch 6/500
Epoch 00006: val_loss did not improve from 0.00554
17/17 - 0s - loss: 0.0210 - accuracy: 0.0000e+00 - val_loss: 0.0453 - val_accuracy: 0.0037 - lr: 0.0010 - 103ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.00554
17/17 - 0s - loss: 0.0146 - accuracy: 0.0000e+00 - val_loss: 0.0367 - val_accuracy: 0.0037 - lr: 0.0010 - 108ms/epoch - 6ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.00554
17/17 - 0s - loss: 0.0165 - accuracy: 0.0000e+00 - val_loss: 0.0063 - val_accuracy: 0.0037 - lr: 0.0010 - 103ms/epoch - 6ms/step
Epoch 9/500
Epoch 00009: val_loss improved from 0.00554 to 0.00500, saving model to LSTM2.h5
17/17 - 0s - loss: 0.0169 - accuracy: 0.0000e+00 - val_loss: 0.0050 - val_accuracy: 0.0037 - lr: 0.0010 - 135ms/epoch - 8ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.00500
17/17 - 0s - loss: 0.0069 - accuracy: 0.0000e+00 - val_loss: 0.0310 - val_accuracy: 0.0037 - lr: 0.0010 - 99ms/epoch - 6ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.00500
17/17 - 0s - loss: 0.0106 - accuracy: 0.0000e+00 - val_loss: 0.0083 - val_accuracy: 0.0037 - lr: 0.0010 - 101ms/epoch - 6ms/step
Epoch 12/500
Epoch 00012: val_loss improved from 0.00500 to 0.00478, saving model to LSTM2.h5
17/17 - 0s - loss: 0.0041 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 0.0010 - 123ms/epoch - 7ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.00478
17/17 - 0s - loss: 0.0083 - accuracy: 0.0000e+00 - val_loss: 0.0088 - val_accuracy: 0.0037 - lr: 0.0010 - 114ms/epoch - 7ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.00478
17/17 - 0s - loss: 0.0022 - accuracy: 0.0000e+00 - val_loss: 0.0157 - val_accuracy: 0.0037 - lr: 0.0010 - 107ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: val_loss improved from 0.00478 to 0.00474, saving model to LSTM2.h5
17/17 - 0s - loss: 0.0050 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 0.0010 - 126ms/epoch - 7ms/step
Epoch 16/500
Epoch 00016: val_loss improved from 0.00474 to 0.00383, saving model to LSTM2.h5
17/17 - 0s - loss: 0.0026 - accuracy: 0.0000e+00 - val_loss: 0.0038 - val_accuracy: 0.0037 - lr: 0.0010 - 124ms/epoch - 7ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.00383
17/17 - 0s - loss: 0.0042 - accuracy: 0.0000e+00 - val_loss: 0.0091 - val_accuracy: 0.0037 - lr: 0.0010 - 104ms/epoch - 6ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.00383
17/17 - 0s - loss: 0.0018 - accuracy: 0.0000e+00 - val_loss: 0.0122 - val_accuracy: 0.0037 - lr: 0.0010 - 119ms/epoch - 7ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.00383
17/17 - 0s - loss: 0.0035 - accuracy: 0.0000e+00 - val_loss: 0.0053 - val_accuracy: 0.0037 - lr: 0.0010 - 132ms/epoch - 8ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.00383
17/17 - 0s - loss: 0.0023 - accuracy: 0.0000e+00 - val_loss: 0.0039 - val_accuracy: 0.0037 - lr: 0.0010 - 102ms/epoch - 6ms/step
Epoch 21/500
Epoch 00021: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00021: val_loss did not improve from 0.00383
17/17 - 0s - loss: 0.0040 - accuracy: 0.0000e+00 - val_loss: 0.0080 - val_accuracy: 0.0037 - lr: 0.0010 - 125ms/epoch - 7ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.00383
17/17 - 0s - loss: 0.0018 - accuracy: 0.0000e+00 - val_loss: 0.0082 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 110ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.00383
17/17 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0083 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 107ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.00383
17/17 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0077 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 120ms/epoch - 7ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.00383
17/17 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0071 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 98ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00026: val_loss did not improve from 0.00383
17/17 - 0s - loss: 9.5297e-04 - accuracy: 0.0000e+00 - val_loss: 0.0068 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 117ms/epoch - 7ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.00383
17/17 - 0s - loss: 9.0436e-04 - accuracy: 0.0000e+00 - val_loss: 0.0068 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 107ms/epoch - 6ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.00383
17/17 - 0s - loss: 9.0057e-04 - accuracy: 0.0000e+00 - val_loss: 0.0067 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.9689e-04 - accuracy: 0.0000e+00 - val_loss: 0.0067 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 105ms/epoch - 6ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.9330e-04 - accuracy: 0.0000e+00 - val_loss: 0.0067 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 108ms/epoch - 6ms/step
Epoch 31/500
Epoch 00031: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00031: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.8980e-04 - accuracy: 0.0000e+00 - val_loss: 0.0067 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 108ms/epoch - 6ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.8639e-04 - accuracy: 0.0000e+00 - val_loss: 0.0067 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 108ms/epoch - 6ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.8308e-04 - accuracy: 0.0000e+00 - val_loss: 0.0066 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 110ms/epoch - 6ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.7985e-04 - accuracy: 0.0000e+00 - val_loss: 0.0066 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 113ms/epoch - 7ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.7672e-04 - accuracy: 0.0000e+00 - val_loss: 0.0066 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 109ms/epoch - 6ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.7368e-04 - accuracy: 0.0000e+00 - val_loss: 0.0066 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 100ms/epoch - 6ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.7074e-04 - accuracy: 0.0000e+00 - val_loss: 0.0066 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 113ms/epoch - 7ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.6788e-04 - accuracy: 0.0000e+00 - val_loss: 0.0066 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.6512e-04 - accuracy: 0.0000e+00 - val_loss: 0.0066 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 118ms/epoch - 7ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.6245e-04 - accuracy: 0.0000e+00 - val_loss: 0.0065 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.5987e-04 - accuracy: 0.0000e+00 - val_loss: 0.0065 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 109ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.5738e-04 - accuracy: 0.0000e+00 - val_loss: 0.0065 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 110ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.5498e-04 - accuracy: 0.0000e+00 - val_loss: 0.0065 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.5266e-04 - accuracy: 0.0000e+00 - val_loss: 0.0065 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.5043e-04 - accuracy: 0.0000e+00 - val_loss: 0.0065 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 108ms/epoch - 6ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.4828e-04 - accuracy: 0.0000e+00 - val_loss: 0.0065 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 105ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.4622e-04 - accuracy: 0.0000e+00 - val_loss: 0.0065 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 106ms/epoch - 6ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.4423e-04 - accuracy: 0.0000e+00 - val_loss: 0.0065 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 115ms/epoch - 7ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.4232e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 112ms/epoch - 7ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.4049e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.3873e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 114ms/epoch - 7ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.3704e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 106ms/epoch - 6ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.3542e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.3387e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.3238e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 106ms/epoch - 6ms/step
Epoch 56/500
Epoch 00056: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.3095e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 105ms/epoch - 6ms/step
Epoch 57/500
Epoch 00057: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.2958e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 122ms/epoch - 7ms/step
Epoch 58/500
Epoch 00058: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.2827e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 124ms/epoch - 7ms/step
Epoch 59/500
Epoch 00059: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.2701e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 124ms/epoch - 7ms/step
Epoch 60/500
Epoch 00060: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.2580e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 121ms/epoch - 7ms/step
Epoch 61/500
Epoch 00061: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.2465e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 114ms/epoch - 7ms/step
Epoch 62/500
Epoch 00062: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.2354e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 104ms/epoch - 6ms/step
Epoch 63/500
Epoch 00063: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.2247e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step
Epoch 64/500
Epoch 00064: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.2145e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 104ms/epoch - 6ms/step
Epoch 65/500
Epoch 00065: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.2047e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 110ms/epoch - 6ms/step
Epoch 66/500
Epoch 00066: val_loss did not improve from 0.00383
17/17 - 0s - loss: 8.1953e-04 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 112ms/epoch - 7ms/step
Epoch 00066: early stopping
SMA
Prediction vs Close: 53.73% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 39.87524298067742
RMSE: 6.3146847095225125
MAPE: 5.088204561858354
EMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 43.66% Accuracy
MSE: 73.6938994303845
RMSE: 8.584515095821342
MAPE: 7.207683534137507
WMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 76.90233333404232
RMSE: 8.76939754681257
MAPE: 7.121360770950225
DEMA
DEMA([input_arrays], [timeperiod=30])
Double Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
89
Working on DEMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.42 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4436.126, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3965.317, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.38 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3736.589, Time=0.06 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3598.951, Time=0.09 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=0.97 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.94 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3600.951, Time=0.20 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.128 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1795.475
Date: Sun, 12 Dec 2021 AIC 3598.951
Time: 14:04:40 BIC 3617.714
Sample: 0 HQIC 3606.157
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1983 0.003 -389.581 0.000 -1.204 -1.192
ar.L2 -0.8973 0.006 -139.732 0.000 -0.910 -0.885
ar.L3 -0.3983 0.006 -68.649 0.000 -0.410 -0.387
sigma2 5.0573 0.023 215.292 0.000 5.011 5.103
===================================================================================
Ljung-Box (L1) (Q): 14.41 Jarque-Bera (JB): 2460553.80
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.89
Prob(H) (two-sided): 0.00 Kurtosis: 273.74
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.05011, saving model to LSTM2.h5
10/10 - 4s - loss: 0.1439 - accuracy: 0.0000e+00 - val_loss: 0.0501 - val_accuracy: 0.0037 - lr: 0.0010 - 4s/epoch - 412ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.05011 to 0.00966, saving model to LSTM2.h5
10/10 - 0s - loss: 0.1247 - accuracy: 0.0000e+00 - val_loss: 0.0097 - val_accuracy: 0.0037 - lr: 0.0010 - 98ms/epoch - 10ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0865 - accuracy: 0.0000e+00 - val_loss: 0.1573 - val_accuracy: 0.0000e+00 - lr: 0.0010 - 73ms/epoch - 7ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0390 - accuracy: 0.0000e+00 - val_loss: 0.0470 - val_accuracy: 0.0037 - lr: 0.0010 - 74ms/epoch - 7ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0124 - accuracy: 0.0000e+00 - val_loss: 0.0357 - val_accuracy: 0.0037 - lr: 0.0010 - 67ms/epoch - 7ms/step
Epoch 6/500
Epoch 00006: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0035 - accuracy: 0.0000e+00 - val_loss: 0.0244 - val_accuracy: 0.0037 - lr: 0.0010 - 71ms/epoch - 7ms/step
Epoch 7/500
Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00007: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0018 - accuracy: 0.0000e+00 - val_loss: 0.0171 - val_accuracy: 0.0037 - lr: 0.0010 - 73ms/epoch - 7ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0018 - accuracy: 0.0000e+00 - val_loss: 0.0183 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 69ms/epoch - 7ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0016 - accuracy: 0.0000e+00 - val_loss: 0.0200 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 75ms/epoch - 8ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0214 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 69ms/epoch - 7ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0222 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 76ms/epoch - 8ms/step
Epoch 12/500
Epoch 00012: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00012: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0225 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 77ms/epoch - 8ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0226 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 81ms/epoch - 8ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0226 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 69ms/epoch - 7ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0226 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 86ms/epoch - 9ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0226 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 70ms/epoch - 7ms/step
Epoch 17/500
Epoch 00017: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00017: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0226 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 81ms/epoch - 8ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0226 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 89ms/epoch - 9ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0226 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 83ms/epoch - 8ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0227 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 69ms/epoch - 7ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0227 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 71ms/epoch - 7ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0227 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 66ms/epoch - 7ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0227 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 72ms/epoch - 7ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0228 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 73ms/epoch - 7ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0228 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 83ms/epoch - 8ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0228 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 75ms/epoch - 7ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0229 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 84ms/epoch - 8ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0229 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 80ms/epoch - 8ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0229 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 74ms/epoch - 7ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0229 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 87ms/epoch - 9ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0230 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 69ms/epoch - 7ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0230 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 73ms/epoch - 7ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0230 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 70ms/epoch - 7ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0231 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 74ms/epoch - 7ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0231 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 70ms/epoch - 7ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0231 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 75ms/epoch - 7ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0231 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 90ms/epoch - 9ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0232 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 78ms/epoch - 8ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0232 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 81ms/epoch - 8ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0232 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 90ms/epoch - 9ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0233 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 85ms/epoch - 9ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0233 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 73ms/epoch - 7ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0233 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 97ms/epoch - 10ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0233 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 66ms/epoch - 7ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0234 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 84ms/epoch - 8ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0234 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 71ms/epoch - 7ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0234 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 66ms/epoch - 7ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0234 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 70ms/epoch - 7ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0234 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 83ms/epoch - 8ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0235 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 76ms/epoch - 8ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0235 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 77ms/epoch - 8ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.00966
10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0235 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 81ms/epoch - 8ms/step
Epoch 00052: early stopping
SMA
Prediction vs Close: 53.73% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 39.87524298067742
RMSE: 6.3146847095225125
MAPE: 5.088204561858354
EMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 43.66% Accuracy
MSE: 73.6938994303845
RMSE: 8.584515095821342
MAPE: 7.207683534137507
WMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 76.90233333404232
RMSE: 8.76939754681257
MAPE: 7.121360770950225
DEMA
Prediction vs Close: 52.24% Accuracy
Prediction vs Prediction: 46.27% Accuracy
MSE: 131.0602137141292
RMSE: 11.448153288374908
MAPE: 10.329401784343453
KAMA
KAMA([input_arrays], [timeperiod=30])
Kaufman Adaptive Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
18
Working on KAMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.36 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4190.464, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3724.371, Time=0.05 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.29 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3494.154, Time=0.09 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3357.435, Time=0.13 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.28 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.78 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3359.435, Time=0.19 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.204 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1674.717
Date: Sun, 12 Dec 2021 AIC 3357.435
Time: 14:05:53 BIC 3376.198
Sample: 0 HQIC 3364.641
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1955 0.003 -381.246 0.000 -1.202 -1.189
ar.L2 -0.8964 0.007 -135.835 0.000 -0.909 -0.883
ar.L3 -0.3971 0.006 -67.229 0.000 -0.409 -0.385
sigma2 3.7466 0.018 211.623 0.000 3.712 3.781
===================================================================================
Ljung-Box (L1) (Q): 14.20 Jarque-Bera (JB): 2338363.32
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.01 Skew: 3.76
Prob(H) (two-sided): 0.00 Kurtosis: 266.93
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.15923, saving model to LSTM2.h5
45/45 - 4s - loss: 0.1495 - accuracy: 0.0000e+00 - val_loss: 0.1592 - val_accuracy: 0.0000e+00 - lr: 0.0010 - 4s/epoch - 86ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.15923 to 0.01635, saving model to LSTM2.h5
45/45 - 0s - loss: 0.0565 - accuracy: 0.0000e+00 - val_loss: 0.0164 - val_accuracy: 0.0037 - lr: 0.0010 - 322ms/epoch - 7ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.01635
45/45 - 0s - loss: 0.0149 - accuracy: 0.0000e+00 - val_loss: 0.0477 - val_accuracy: 0.0037 - lr: 0.0010 - 255ms/epoch - 6ms/step
Epoch 4/500
Epoch 00004: val_loss improved from 0.01635 to 0.00499, saving model to LSTM2.h5
45/45 - 0s - loss: 0.0160 - accuracy: 0.0000e+00 - val_loss: 0.0050 - val_accuracy: 0.0037 - lr: 0.0010 - 285ms/epoch - 6ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.00499
45/45 - 0s - loss: 0.0066 - accuracy: 0.0000e+00 - val_loss: 0.0185 - val_accuracy: 0.0037 - lr: 0.0010 - 253ms/epoch - 6ms/step
Epoch 6/500
Epoch 00006: val_loss improved from 0.00499 to 0.00365, saving model to LSTM2.h5
45/45 - 0s - loss: 0.0046 - accuracy: 0.0000e+00 - val_loss: 0.0036 - val_accuracy: 0.0037 - lr: 0.0010 - 270ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.00365
45/45 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0065 - val_accuracy: 0.0037 - lr: 0.0010 - 237ms/epoch - 5ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.00365
45/45 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 0.0010 - 238ms/epoch - 5ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.00365
45/45 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 0.0010 - 256ms/epoch - 6ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.00365
45/45 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0063 - val_accuracy: 0.0037 - lr: 0.0010 - 292ms/epoch - 6ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00011: val_loss did not improve from 0.00365
45/45 - 0s - loss: 0.0018 - accuracy: 0.0000e+00 - val_loss: 0.0091 - val_accuracy: 0.0037 - lr: 0.0010 - 280ms/epoch - 6ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.00365
45/45 - 0s - loss: 0.0024 - accuracy: 0.0000e+00 - val_loss: 0.0052 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 259ms/epoch - 6ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.00365
45/45 - 0s - loss: 0.0025 - accuracy: 0.0000e+00 - val_loss: 0.0074 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 251ms/epoch - 6ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.00365
45/45 - 0s - loss: 0.0015 - accuracy: 0.0000e+00 - val_loss: 0.0085 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 282ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.00365
45/45 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0095 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 273ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00016: val_loss did not improve from 0.00365
45/45 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0102 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 270ms/epoch - 6ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.00365
45/45 - 0s - loss: 9.0282e-04 - accuracy: 0.0000e+00 - val_loss: 0.0104 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 249ms/epoch - 6ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.8947e-04 - accuracy: 0.0000e+00 - val_loss: 0.0105 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 257ms/epoch - 6ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.8125e-04 - accuracy: 0.0000e+00 - val_loss: 0.0106 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 273ms/epoch - 6ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.7539e-04 - accuracy: 0.0000e+00 - val_loss: 0.0107 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 234ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00021: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.7045e-04 - accuracy: 0.0000e+00 - val_loss: 0.0108 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 295ms/epoch - 7ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.6589e-04 - accuracy: 0.0000e+00 - val_loss: 0.0108 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 272ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.6154e-04 - accuracy: 0.0000e+00 - val_loss: 0.0109 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 255ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.5733e-04 - accuracy: 0.0000e+00 - val_loss: 0.0109 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 233ms/epoch - 5ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.5323e-04 - accuracy: 0.0000e+00 - val_loss: 0.0110 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 287ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.4924e-04 - accuracy: 0.0000e+00 - val_loss: 0.0111 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 245ms/epoch - 5ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.4535e-04 - accuracy: 0.0000e+00 - val_loss: 0.0111 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 296ms/epoch - 7ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.4156e-04 - accuracy: 0.0000e+00 - val_loss: 0.0112 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 262ms/epoch - 6ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.3788e-04 - accuracy: 0.0000e+00 - val_loss: 0.0112 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 290ms/epoch - 6ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.3430e-04 - accuracy: 0.0000e+00 - val_loss: 0.0113 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 272ms/epoch - 6ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.3082e-04 - accuracy: 0.0000e+00 - val_loss: 0.0114 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 289ms/epoch - 6ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.2745e-04 - accuracy: 0.0000e+00 - val_loss: 0.0114 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 291ms/epoch - 6ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.2418e-04 - accuracy: 0.0000e+00 - val_loss: 0.0115 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 309ms/epoch - 7ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.2101e-04 - accuracy: 0.0000e+00 - val_loss: 0.0116 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 269ms/epoch - 6ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.1794e-04 - accuracy: 0.0000e+00 - val_loss: 0.0117 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 272ms/epoch - 6ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.1497e-04 - accuracy: 0.0000e+00 - val_loss: 0.0117 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 251ms/epoch - 6ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.1210e-04 - accuracy: 0.0000e+00 - val_loss: 0.0118 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 252ms/epoch - 6ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.0933e-04 - accuracy: 0.0000e+00 - val_loss: 0.0119 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 236ms/epoch - 5ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.0665e-04 - accuracy: 0.0000e+00 - val_loss: 0.0120 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 264ms/epoch - 6ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.0406e-04 - accuracy: 0.0000e+00 - val_loss: 0.0120 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 283ms/epoch - 6ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.00365
45/45 - 0s - loss: 8.0157e-04 - accuracy: 0.0000e+00 - val_loss: 0.0121 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 276ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.00365
45/45 - 0s - loss: 7.9917e-04 - accuracy: 0.0000e+00 - val_loss: 0.0122 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 276ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.00365
45/45 - 0s - loss: 7.9686e-04 - accuracy: 0.0000e+00 - val_loss: 0.0123 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 255ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.00365
45/45 - 0s - loss: 7.9464e-04 - accuracy: 0.0000e+00 - val_loss: 0.0124 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 262ms/epoch - 6ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.00365
45/45 - 0s - loss: 7.9250e-04 - accuracy: 0.0000e+00 - val_loss: 0.0124 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 284ms/epoch - 6ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.00365
45/45 - 0s - loss: 7.9045e-04 - accuracy: 0.0000e+00 - val_loss: 0.0125 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 241ms/epoch - 5ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.00365
45/45 - 0s - loss: 7.8848e-04 - accuracy: 0.0000e+00 - val_loss: 0.0126 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 261ms/epoch - 6ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.00365
45/45 - 0s - loss: 7.8659e-04 - accuracy: 0.0000e+00 - val_loss: 0.0127 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 241ms/epoch - 5ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.00365
45/45 - 0s - loss: 7.8477e-04 - accuracy: 0.0000e+00 - val_loss: 0.0128 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 240ms/epoch - 5ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.00365
45/45 - 0s - loss: 7.8303e-04 - accuracy: 0.0000e+00 - val_loss: 0.0128 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 270ms/epoch - 6ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.00365
45/45 - 0s - loss: 7.8136e-04 - accuracy: 0.0000e+00 - val_loss: 0.0129 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 279ms/epoch - 6ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.00365
45/45 - 0s - loss: 7.7977e-04 - accuracy: 0.0000e+00 - val_loss: 0.0130 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 279ms/epoch - 6ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.00365
45/45 - 0s - loss: 7.7824e-04 - accuracy: 0.0000e+00 - val_loss: 0.0131 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 265ms/epoch - 6ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.00365
45/45 - 0s - loss: 7.7678e-04 - accuracy: 0.0000e+00 - val_loss: 0.0132 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 282ms/epoch - 6ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.00365
45/45 - 0s - loss: 7.7538e-04 - accuracy: 0.0000e+00 - val_loss: 0.0132 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 248ms/epoch - 6ms/step
Epoch 56/500
Epoch 00056: val_loss did not improve from 0.00365
45/45 - 0s - loss: 7.7404e-04 - accuracy: 0.0000e+00 - val_loss: 0.0133 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 276ms/epoch - 6ms/step
Epoch 00056: early stopping
SMA
Prediction vs Close: 53.73% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 39.87524298067742
RMSE: 6.3146847095225125
MAPE: 5.088204561858354
EMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 43.66% Accuracy
MSE: 73.6938994303845
RMSE: 8.584515095821342
MAPE: 7.207683534137507
WMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 76.90233333404232
RMSE: 8.76939754681257
MAPE: 7.121360770950225
DEMA
Prediction vs Close: 52.24% Accuracy
Prediction vs Prediction: 46.27% Accuracy
MSE: 131.0602137141292
RMSE: 11.448153288374908
MAPE: 10.329401784343453
KAMA
Prediction vs Close: 53.36% Accuracy
Prediction vs Prediction: 47.76% Accuracy
MSE: 42.128331463904196
RMSE: 6.49063413418937
MAPE: 5.225311201416733
MIDPOINT
MIDPOINT([input_arrays], [timeperiod=14])
MidPoint over period (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 14
Outputs:
real
14
Working on MIDPOINT predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.36 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4212.289, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3747.746, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.23 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3523.401, Time=0.07 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3387.759, Time=0.09 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.22 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.85 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3389.758, Time=0.22 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.118 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1689.879
Date: Sun, 12 Dec 2021 AIC 3387.759
Time: 14:07:24 BIC 3406.522
Sample: 0 HQIC 3394.964
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1878 0.003 -345.315 0.000 -1.195 -1.181
ar.L2 -0.8876 0.007 -121.809 0.000 -0.902 -0.873
ar.L3 -0.3957 0.007 -60.127 0.000 -0.409 -0.383
sigma2 3.8904 0.020 193.404 0.000 3.851 3.930
===================================================================================
Ljung-Box (L1) (Q): 13.21 Jarque-Bera (JB): 1659080.01
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.08 Skew: 3.28
Prob(H) (two-sided): 0.00 Kurtosis: 225.31
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.01067, saving model to LSTM2.h5
58/58 - 4s - loss: 0.2600 - accuracy: 0.0000e+00 - val_loss: 0.0107 - val_accuracy: 0.0037 - lr: 0.0010 - 4s/epoch - 70ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.01067
58/58 - 0s - loss: 0.0424 - accuracy: 0.0000e+00 - val_loss: 0.0374 - val_accuracy: 0.0037 - lr: 0.0010 - 333ms/epoch - 6ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.01067
58/58 - 0s - loss: 0.0045 - accuracy: 0.0000e+00 - val_loss: 0.0623 - val_accuracy: 0.0037 - lr: 0.0010 - 343ms/epoch - 6ms/step
Epoch 4/500
Epoch 00004: val_loss improved from 0.01067 to 0.00479, saving model to LSTM2.h5
58/58 - 0s - loss: 0.0145 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 0.0010 - 351ms/epoch - 6ms/step
Epoch 5/500
Epoch 00005: val_loss improved from 0.00479 to 0.00428, saving model to LSTM2.h5
58/58 - 0s - loss: 0.0033 - accuracy: 0.0000e+00 - val_loss: 0.0043 - val_accuracy: 0.0037 - lr: 0.0010 - 341ms/epoch - 6ms/step
Epoch 6/500
Epoch 00006: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0046 - accuracy: 0.0000e+00 - val_loss: 0.0049 - val_accuracy: 0.0037 - lr: 0.0010 - 322ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0042 - accuracy: 0.0000e+00 - val_loss: 0.0057 - val_accuracy: 0.0037 - lr: 0.0010 - 340ms/epoch - 6ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0074 - accuracy: 0.0000e+00 - val_loss: 0.0054 - val_accuracy: 0.0037 - lr: 0.0010 - 298ms/epoch - 5ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0154 - accuracy: 0.0000e+00 - val_loss: 0.0057 - val_accuracy: 0.0037 - lr: 0.0010 - 315ms/epoch - 5ms/step
Epoch 10/500
Epoch 00010: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00010: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0342 - accuracy: 0.0000e+00 - val_loss: 0.0071 - val_accuracy: 0.0037 - lr: 0.0010 - 338ms/epoch - 6ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0489 - accuracy: 0.0000e+00 - val_loss: 0.0907 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 307ms/epoch - 5ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0171 - accuracy: 0.0000e+00 - val_loss: 0.0417 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 305ms/epoch - 5ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0096 - accuracy: 0.0000e+00 - val_loss: 0.0307 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 330ms/epoch - 6ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0068 - accuracy: 0.0000e+00 - val_loss: 0.0230 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 321ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00015: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0048 - accuracy: 0.0000e+00 - val_loss: 0.0181 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 316ms/epoch - 5ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0034 - accuracy: 0.0000e+00 - val_loss: 0.0172 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 343ms/epoch - 6ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0032 - accuracy: 0.0000e+00 - val_loss: 0.0165 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 314ms/epoch - 5ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0031 - accuracy: 0.0000e+00 - val_loss: 0.0159 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 317ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0030 - accuracy: 0.0000e+00 - val_loss: 0.0154 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 354ms/epoch - 6ms/step
Epoch 20/500
Epoch 00020: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00020: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0029 - accuracy: 0.0000e+00 - val_loss: 0.0149 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 322ms/epoch - 6ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0028 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 341ms/epoch - 6ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0027 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 337ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0027 - accuracy: 0.0000e+00 - val_loss: 0.0138 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 307ms/epoch - 5ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0026 - accuracy: 0.0000e+00 - val_loss: 0.0135 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 348ms/epoch - 6ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0025 - accuracy: 0.0000e+00 - val_loss: 0.0132 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 328ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0024 - accuracy: 0.0000e+00 - val_loss: 0.0129 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 316ms/epoch - 5ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0024 - accuracy: 0.0000e+00 - val_loss: 0.0125 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 322ms/epoch - 6ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0023 - accuracy: 0.0000e+00 - val_loss: 0.0122 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 330ms/epoch - 6ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0022 - accuracy: 0.0000e+00 - val_loss: 0.0119 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 302ms/epoch - 5ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0021 - accuracy: 0.0000e+00 - val_loss: 0.0116 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 322ms/epoch - 6ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0021 - accuracy: 0.0000e+00 - val_loss: 0.0113 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 363ms/epoch - 6ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0020 - accuracy: 0.0000e+00 - val_loss: 0.0110 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 323ms/epoch - 6ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0019 - accuracy: 0.0000e+00 - val_loss: 0.0107 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 324ms/epoch - 6ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0019 - accuracy: 0.0000e+00 - val_loss: 0.0104 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 366ms/epoch - 6ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0018 - accuracy: 0.0000e+00 - val_loss: 0.0101 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 323ms/epoch - 6ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0018 - accuracy: 0.0000e+00 - val_loss: 0.0098 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 318ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0017 - accuracy: 0.0000e+00 - val_loss: 0.0095 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 313ms/epoch - 5ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0017 - accuracy: 0.0000e+00 - val_loss: 0.0092 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 307ms/epoch - 5ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0016 - accuracy: 0.0000e+00 - val_loss: 0.0089 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 312ms/epoch - 5ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0016 - accuracy: 0.0000e+00 - val_loss: 0.0086 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 346ms/epoch - 6ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0015 - accuracy: 0.0000e+00 - val_loss: 0.0083 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 334ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0015 - accuracy: 0.0000e+00 - val_loss: 0.0080 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 338ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0078 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 339ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0075 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 322ms/epoch - 6ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0073 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 312ms/epoch - 5ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0070 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 321ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0068 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 316ms/epoch - 5ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0065 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 353ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0063 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 393ms/epoch - 7ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0061 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 345ms/epoch - 6ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0059 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 320ms/epoch - 6ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0057 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 325ms/epoch - 6ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0055 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 299ms/epoch - 5ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0054 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 403ms/epoch - 7ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.00428
58/58 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0052 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 339ms/epoch - 6ms/step
Epoch 00055: early stopping
SMA
Prediction vs Close: 53.73% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 39.87524298067742
RMSE: 6.3146847095225125
MAPE: 5.088204561858354
EMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 43.66% Accuracy
MSE: 73.6938994303845
RMSE: 8.584515095821342
MAPE: 7.207683534137507
WMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 76.90233333404232
RMSE: 8.76939754681257
MAPE: 7.121360770950225
DEMA
Prediction vs Close: 52.24% Accuracy
Prediction vs Prediction: 46.27% Accuracy
MSE: 131.0602137141292
RMSE: 11.448153288374908
MAPE: 10.329401784343453
KAMA
Prediction vs Close: 53.36% Accuracy
Prediction vs Prediction: 47.76% Accuracy
MSE: 42.128331463904196
RMSE: 6.49063413418937
MAPE: 5.225311201416733
MIDPOINT
Prediction vs Close: 52.24% Accuracy
Prediction vs Prediction: 45.15% Accuracy
MSE: 102.72361049298229
RMSE: 10.135265684380567
MAPE: 8.372939983805384
T3
T3([input_arrays], [timeperiod=5], [vfactor=0.7])
Triple Exponential Moving Average (T3) (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 5
vfactor: 0.7
Outputs:
real
19
Working on T3 predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.35 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4414.515, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3944.062, Time=0.05 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.38 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3715.173, Time=0.07 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3577.471, Time=0.09 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.49 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.65 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3579.471, Time=0.18 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.292 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1784.736
Date: Sun, 12 Dec 2021 AIC 3577.471
Time: 14:09:00 BIC 3596.235
Sample: 0 HQIC 3584.677
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1982 0.003 -389.844 0.000 -1.204 -1.192
ar.L2 -0.8974 0.006 -139.861 0.000 -0.910 -0.885
ar.L3 -0.3983 0.006 -68.862 0.000 -0.410 -0.387
sigma2 4.9242 0.023 215.469 0.000 4.879 4.969
===================================================================================
Ljung-Box (L1) (Q): 14.55 Jarque-Bera (JB): 2468024.38
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.90
Prob(H) (two-sided): 0.00 Kurtosis: 274.15
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.07207, saving model to LSTM2.h5
43/43 - 5s - loss: 0.1415 - accuracy: 0.0000e+00 - val_loss: 0.0721 - val_accuracy: 0.0037 - lr: 0.0010 - 5s/epoch - 111ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.07207 to 0.03513, saving model to LSTM2.h5
43/43 - 0s - loss: 0.0578 - accuracy: 0.0000e+00 - val_loss: 0.0351 - val_accuracy: 0.0037 - lr: 0.0010 - 243ms/epoch - 6ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.03513
43/43 - 0s - loss: 0.0359 - accuracy: 0.0000e+00 - val_loss: 0.0626 - val_accuracy: 0.0037 - lr: 0.0010 - 220ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss improved from 0.03513 to 0.00755, saving model to LSTM2.h5
43/43 - 0s - loss: 0.0163 - accuracy: 0.0000e+00 - val_loss: 0.0076 - val_accuracy: 0.0037 - lr: 0.0010 - 245ms/epoch - 6ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.00755
43/43 - 0s - loss: 0.0112 - accuracy: 0.0000e+00 - val_loss: 0.0190 - val_accuracy: 0.0037 - lr: 0.0010 - 232ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: val_loss improved from 0.00755 to 0.00438, saving model to LSTM2.h5
43/43 - 0s - loss: 0.0046 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 0.0010 - 272ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.00438
43/43 - 0s - loss: 0.0037 - accuracy: 0.0000e+00 - val_loss: 0.0093 - val_accuracy: 0.0037 - lr: 0.0010 - 281ms/epoch - 7ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.00438
43/43 - 0s - loss: 0.0016 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 0.0010 - 279ms/epoch - 6ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.00438
43/43 - 0s - loss: 0.0016 - accuracy: 0.0000e+00 - val_loss: 0.0067 - val_accuracy: 0.0037 - lr: 0.0010 - 241ms/epoch - 6ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.00438
43/43 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0063 - val_accuracy: 0.0037 - lr: 0.0010 - 280ms/epoch - 7ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00011: val_loss did not improve from 0.00438
43/43 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0075 - val_accuracy: 0.0037 - lr: 0.0010 - 239ms/epoch - 6ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.00438
43/43 - 0s - loss: 0.0018 - accuracy: 0.0000e+00 - val_loss: 0.0089 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 247ms/epoch - 6ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.00438
43/43 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0102 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 267ms/epoch - 6ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.00438
43/43 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0113 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 252ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.00438
43/43 - 0s - loss: 9.7475e-04 - accuracy: 0.0000e+00 - val_loss: 0.0121 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 269ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00016: val_loss did not improve from 0.00438
43/43 - 0s - loss: 9.1016e-04 - accuracy: 0.0000e+00 - val_loss: 0.0128 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 246ms/epoch - 6ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.6487e-04 - accuracy: 0.0000e+00 - val_loss: 0.0129 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 259ms/epoch - 6ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.5760e-04 - accuracy: 0.0000e+00 - val_loss: 0.0130 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 251ms/epoch - 6ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.5277e-04 - accuracy: 0.0000e+00 - val_loss: 0.0131 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 288ms/epoch - 7ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.4948e-04 - accuracy: 0.0000e+00 - val_loss: 0.0132 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 232ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00021: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.4691e-04 - accuracy: 0.0000e+00 - val_loss: 0.0133 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 262ms/epoch - 6ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.4467e-04 - accuracy: 0.0000e+00 - val_loss: 0.0133 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 274ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.4258e-04 - accuracy: 0.0000e+00 - val_loss: 0.0134 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 243ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.4055e-04 - accuracy: 0.0000e+00 - val_loss: 0.0135 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 237ms/epoch - 6ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.3856e-04 - accuracy: 0.0000e+00 - val_loss: 0.0135 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 238ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.3660e-04 - accuracy: 0.0000e+00 - val_loss: 0.0136 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 229ms/epoch - 5ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.3464e-04 - accuracy: 0.0000e+00 - val_loss: 0.0136 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 237ms/epoch - 6ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.3270e-04 - accuracy: 0.0000e+00 - val_loss: 0.0137 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 244ms/epoch - 6ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.3077e-04 - accuracy: 0.0000e+00 - val_loss: 0.0137 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 236ms/epoch - 5ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.2885e-04 - accuracy: 0.0000e+00 - val_loss: 0.0138 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 253ms/epoch - 6ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.2695e-04 - accuracy: 0.0000e+00 - val_loss: 0.0138 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 268ms/epoch - 6ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.2506e-04 - accuracy: 0.0000e+00 - val_loss: 0.0139 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 231ms/epoch - 5ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.2320e-04 - accuracy: 0.0000e+00 - val_loss: 0.0140 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 276ms/epoch - 6ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.2135e-04 - accuracy: 0.0000e+00 - val_loss: 0.0140 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 270ms/epoch - 6ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.1952e-04 - accuracy: 0.0000e+00 - val_loss: 0.0141 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 236ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.1772e-04 - accuracy: 0.0000e+00 - val_loss: 0.0141 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 233ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.1593e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 233ms/epoch - 5ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.1418e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 230ms/epoch - 5ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.1244e-04 - accuracy: 0.0000e+00 - val_loss: 0.0143 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 243ms/epoch - 6ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.1073e-04 - accuracy: 0.0000e+00 - val_loss: 0.0144 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 256ms/epoch - 6ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.0905e-04 - accuracy: 0.0000e+00 - val_loss: 0.0144 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 270ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.0739e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 242ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.0576e-04 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 268ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.0415e-04 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 274ms/epoch - 6ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.0257e-04 - accuracy: 0.0000e+00 - val_loss: 0.0147 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 245ms/epoch - 6ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.00438
43/43 - 0s - loss: 8.0102e-04 - accuracy: 0.0000e+00 - val_loss: 0.0148 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 240ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.00438
43/43 - 0s - loss: 7.9949e-04 - accuracy: 0.0000e+00 - val_loss: 0.0148 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 257ms/epoch - 6ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.00438
43/43 - 0s - loss: 7.9798e-04 - accuracy: 0.0000e+00 - val_loss: 0.0149 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 246ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.00438
43/43 - 0s - loss: 7.9650e-04 - accuracy: 0.0000e+00 - val_loss: 0.0150 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 260ms/epoch - 6ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.00438
43/43 - 0s - loss: 7.9504e-04 - accuracy: 0.0000e+00 - val_loss: 0.0150 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 228ms/epoch - 5ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.00438
43/43 - 0s - loss: 7.9361e-04 - accuracy: 0.0000e+00 - val_loss: 0.0151 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 229ms/epoch - 5ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.00438
43/43 - 0s - loss: 7.9220e-04 - accuracy: 0.0000e+00 - val_loss: 0.0152 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 229ms/epoch - 5ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.00438
43/43 - 0s - loss: 7.9081e-04 - accuracy: 0.0000e+00 - val_loss: 0.0152 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 277ms/epoch - 6ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.00438
43/43 - 0s - loss: 7.8944e-04 - accuracy: 0.0000e+00 - val_loss: 0.0153 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 227ms/epoch - 5ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.00438
43/43 - 0s - loss: 7.8809e-04 - accuracy: 0.0000e+00 - val_loss: 0.0154 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 241ms/epoch - 6ms/step
Epoch 56/500
Epoch 00056: val_loss did not improve from 0.00438
43/43 - 0s - loss: 7.8677e-04 - accuracy: 0.0000e+00 - val_loss: 0.0154 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 260ms/epoch - 6ms/step
Epoch 00056: early stopping
SMA
Prediction vs Close: 53.73% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 39.87524298067742
RMSE: 6.3146847095225125
MAPE: 5.088204561858354
EMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 43.66% Accuracy
MSE: 73.6938994303845
RMSE: 8.584515095821342
MAPE: 7.207683534137507
WMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 76.90233333404232
RMSE: 8.76939754681257
MAPE: 7.121360770950225
DEMA
Prediction vs Close: 52.24% Accuracy
Prediction vs Prediction: 46.27% Accuracy
MSE: 131.0602137141292
RMSE: 11.448153288374908
MAPE: 10.329401784343453
KAMA
Prediction vs Close: 53.36% Accuracy
Prediction vs Prediction: 47.76% Accuracy
MSE: 42.128331463904196
RMSE: 6.49063413418937
MAPE: 5.225311201416733
MIDPOINT
Prediction vs Close: 52.24% Accuracy
Prediction vs Prediction: 45.15% Accuracy
MSE: 102.72361049298229
RMSE: 10.135265684380567
MAPE: 8.372939983805384
T3
Prediction vs Close: 55.22% Accuracy
Prediction vs Prediction: 45.15% Accuracy
MSE: 106.78678832331099
RMSE: 10.333769318274479
MAPE: 8.334873887974807
TEMA
TEMA([input_arrays], [timeperiod=30])
Triple Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
9
Working on TEMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.45 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4352.703, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3889.412, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.26 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3689.930, Time=0.06 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3574.245, Time=0.08 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.10 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.75 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3576.245, Time=0.17 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 2.950 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1783.123
Date: Sun, 12 Dec 2021 AIC 3574.245
Time: 14:10:25 BIC 3593.008
Sample: 0 HQIC 3581.451
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1480 0.004 -302.430 0.000 -1.155 -1.141
ar.L2 -0.8300 0.008 -99.682 0.000 -0.846 -0.814
ar.L3 -0.3687 0.007 -50.527 0.000 -0.383 -0.354
sigma2 4.9055 0.028 175.970 0.000 4.851 4.960
===================================================================================
Ljung-Box (L1) (Q): 11.61 Jarque-Bera (JB): 1261976.58
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.16 Skew: 2.52
Prob(H) (two-sided): 0.00 Kurtosis: 196.90
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.12996, saving model to LSTM2.h5
90/90 - 4s - loss: 0.1545 - accuracy: 0.0000e+00 - val_loss: 0.1300 - val_accuracy: 0.0000e+00 - lr: 0.0010 - 4s/epoch - 49ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.12996
90/90 - 0s - loss: 0.1665 - accuracy: 0.0000e+00 - val_loss: 0.2427 - val_accuracy: 0.0000e+00 - lr: 0.0010 - 462ms/epoch - 5ms/step
Epoch 3/500
Epoch 00003: val_loss improved from 0.12996 to 0.00886, saving model to LSTM2.h5
90/90 - 0s - loss: 0.0386 - accuracy: 0.0000e+00 - val_loss: 0.0089 - val_accuracy: 0.0037 - lr: 0.0010 - 484ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.00886
90/90 - 0s - loss: 0.0097 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 0.0010 - 459ms/epoch - 5ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.00886
90/90 - 0s - loss: 0.0025 - accuracy: 0.0000e+00 - val_loss: 0.0226 - val_accuracy: 0.0037 - lr: 0.0010 - 480ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: val_loss did not improve from 0.00886
90/90 - 0s - loss: 0.0029 - accuracy: 0.0000e+00 - val_loss: 0.0153 - val_accuracy: 0.0037 - lr: 0.0010 - 474ms/epoch - 5ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.00886
90/90 - 1s - loss: 0.0034 - accuracy: 0.0000e+00 - val_loss: 0.0117 - val_accuracy: 0.0037 - lr: 0.0010 - 544ms/epoch - 6ms/step
Epoch 8/500
Epoch 00008: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00008: val_loss did not improve from 0.00886
90/90 - 1s - loss: 0.0041 - accuracy: 0.0000e+00 - val_loss: 0.0175 - val_accuracy: 0.0037 - lr: 0.0010 - 505ms/epoch - 6ms/step
Epoch 9/500
Epoch 00009: val_loss improved from 0.00886 to 0.00880, saving model to LSTM2.h5
90/90 - 1s - loss: 0.0044 - accuracy: 0.0000e+00 - val_loss: 0.0088 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 501ms/epoch - 6ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.00880
90/90 - 0s - loss: 0.0044 - accuracy: 0.0000e+00 - val_loss: 0.0089 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 453ms/epoch - 5ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.00880
90/90 - 0s - loss: 0.0027 - accuracy: 0.0000e+00 - val_loss: 0.0090 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 465ms/epoch - 5ms/step
Epoch 12/500
Epoch 00012: val_loss improved from 0.00880 to 0.00879, saving model to LSTM2.h5
90/90 - 0s - loss: 0.0022 - accuracy: 0.0000e+00 - val_loss: 0.0088 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 463ms/epoch - 5ms/step
Epoch 13/500
Epoch 00013: val_loss improved from 0.00879 to 0.00862, saving model to LSTM2.h5
90/90 - 0s - loss: 0.0019 - accuracy: 0.0000e+00 - val_loss: 0.0086 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 474ms/epoch - 5ms/step
Epoch 14/500
Epoch 00014: val_loss improved from 0.00862 to 0.00858, saving model to LSTM2.h5
90/90 - 1s - loss: 0.0017 - accuracy: 0.0000e+00 - val_loss: 0.0086 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 583ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.00858
90/90 - 1s - loss: 0.0016 - accuracy: 0.0000e+00 - val_loss: 0.0087 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 543ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.00858
90/90 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0089 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 467ms/epoch - 5ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.00858
90/90 - 1s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0093 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 529ms/epoch - 6ms/step
Epoch 18/500
Epoch 00018: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00018: val_loss did not improve from 0.00858
90/90 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0098 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 463ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.00858
90/90 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0101 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 485ms/epoch - 5ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.00858
90/90 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0104 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 476ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.00858
90/90 - 1s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0106 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 556ms/epoch - 6ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.00858
90/90 - 1s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0108 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 521ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00023: val_loss did not improve from 0.00858
90/90 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0110 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 476ms/epoch - 5ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.00858
90/90 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0112 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 456ms/epoch - 5ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.00858
90/90 - 1s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0114 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 546ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.00858
90/90 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0115 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 463ms/epoch - 5ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.00858
90/90 - 0s - loss: 9.9422e-04 - accuracy: 0.0000e+00 - val_loss: 0.0117 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 473ms/epoch - 5ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.00858
90/90 - 1s - loss: 9.8459e-04 - accuracy: 0.0000e+00 - val_loss: 0.0119 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 549ms/epoch - 6ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.00858
90/90 - 0s - loss: 9.7538e-04 - accuracy: 0.0000e+00 - val_loss: 0.0121 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 475ms/epoch - 5ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.00858
90/90 - 0s - loss: 9.6655e-04 - accuracy: 0.0000e+00 - val_loss: 0.0123 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 470ms/epoch - 5ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.00858
90/90 - 1s - loss: 9.5807e-04 - accuracy: 0.0000e+00 - val_loss: 0.0125 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 529ms/epoch - 6ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.00858
90/90 - 0s - loss: 9.4989e-04 - accuracy: 0.0000e+00 - val_loss: 0.0127 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 460ms/epoch - 5ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.00858
90/90 - 1s - loss: 9.4200e-04 - accuracy: 0.0000e+00 - val_loss: 0.0129 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 560ms/epoch - 6ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.00858
90/90 - 0s - loss: 9.3435e-04 - accuracy: 0.0000e+00 - val_loss: 0.0131 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 475ms/epoch - 5ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.00858
90/90 - 0s - loss: 9.2694e-04 - accuracy: 0.0000e+00 - val_loss: 0.0133 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 479ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.00858
90/90 - 0s - loss: 9.1975e-04 - accuracy: 0.0000e+00 - val_loss: 0.0135 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 463ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.00858
90/90 - 1s - loss: 9.1276e-04 - accuracy: 0.0000e+00 - val_loss: 0.0137 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 540ms/epoch - 6ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.00858
90/90 - 1s - loss: 9.0597e-04 - accuracy: 0.0000e+00 - val_loss: 0.0140 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 513ms/epoch - 6ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.00858
90/90 - 1s - loss: 8.9935e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 627ms/epoch - 7ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.00858
90/90 - 0s - loss: 8.9291e-04 - accuracy: 0.0000e+00 - val_loss: 0.0144 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 480ms/epoch - 5ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.00858
90/90 - 0s - loss: 8.8664e-04 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 468ms/epoch - 5ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.00858
90/90 - 0s - loss: 8.8052e-04 - accuracy: 0.0000e+00 - val_loss: 0.0149 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 453ms/epoch - 5ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.00858
90/90 - 0s - loss: 8.7456e-04 - accuracy: 0.0000e+00 - val_loss: 0.0151 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 476ms/epoch - 5ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.00858
90/90 - 0s - loss: 8.6874e-04 - accuracy: 0.0000e+00 - val_loss: 0.0154 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 487ms/epoch - 5ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.00858
90/90 - 0s - loss: 8.6307e-04 - accuracy: 0.0000e+00 - val_loss: 0.0156 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 466ms/epoch - 5ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.00858
90/90 - 1s - loss: 8.5753e-04 - accuracy: 0.0000e+00 - val_loss: 0.0159 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 534ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.00858
90/90 - 0s - loss: 8.5213e-04 - accuracy: 0.0000e+00 - val_loss: 0.0161 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 474ms/epoch - 5ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.00858
90/90 - 1s - loss: 8.4685e-04 - accuracy: 0.0000e+00 - val_loss: 0.0164 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 547ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.00858
90/90 - 0s - loss: 8.4171e-04 - accuracy: 0.0000e+00 - val_loss: 0.0166 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 472ms/epoch - 5ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.00858
90/90 - 1s - loss: 8.3668e-04 - accuracy: 0.0000e+00 - val_loss: 0.0169 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 548ms/epoch - 6ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.00858
90/90 - 0s - loss: 8.3178e-04 - accuracy: 0.0000e+00 - val_loss: 0.0172 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 488ms/epoch - 5ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.00858
90/90 - 1s - loss: 8.2699e-04 - accuracy: 0.0000e+00 - val_loss: 0.0175 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 589ms/epoch - 7ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.00858
90/90 - 1s - loss: 8.2232e-04 - accuracy: 0.0000e+00 - val_loss: 0.0178 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 537ms/epoch - 6ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.00858
90/90 - 0s - loss: 8.1777e-04 - accuracy: 0.0000e+00 - val_loss: 0.0180 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 462ms/epoch - 5ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.00858
90/90 - 1s - loss: 8.1333e-04 - accuracy: 0.0000e+00 - val_loss: 0.0183 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 541ms/epoch - 6ms/step
Epoch 56/500
Epoch 00056: val_loss did not improve from 0.00858
90/90 - 0s - loss: 8.0900e-04 - accuracy: 0.0000e+00 - val_loss: 0.0186 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 455ms/epoch - 5ms/step
Epoch 57/500
Epoch 00057: val_loss did not improve from 0.00858
90/90 - 0s - loss: 8.0478e-04 - accuracy: 0.0000e+00 - val_loss: 0.0189 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 462ms/epoch - 5ms/step
Epoch 58/500
Epoch 00058: val_loss did not improve from 0.00858
90/90 - 0s - loss: 8.0068e-04 - accuracy: 0.0000e+00 - val_loss: 0.0192 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 464ms/epoch - 5ms/step
Epoch 59/500
Epoch 00059: val_loss did not improve from 0.00858
90/90 - 0s - loss: 7.9669e-04 - accuracy: 0.0000e+00 - val_loss: 0.0195 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 455ms/epoch - 5ms/step
Epoch 60/500
Epoch 00060: val_loss did not improve from 0.00858
90/90 - 1s - loss: 7.9282e-04 - accuracy: 0.0000e+00 - val_loss: 0.0198 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 529ms/epoch - 6ms/step
Epoch 61/500
Epoch 00061: val_loss did not improve from 0.00858
90/90 - 1s - loss: 7.8906e-04 - accuracy: 0.0000e+00 - val_loss: 0.0202 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 547ms/epoch - 6ms/step
Epoch 62/500
Epoch 00062: val_loss did not improve from 0.00858
90/90 - 0s - loss: 7.8541e-04 - accuracy: 0.0000e+00 - val_loss: 0.0205 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 476ms/epoch - 5ms/step
Epoch 63/500
Epoch 00063: val_loss did not improve from 0.00858
90/90 - 0s - loss: 7.8187e-04 - accuracy: 0.0000e+00 - val_loss: 0.0208 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 469ms/epoch - 5ms/step
Epoch 64/500
Epoch 00064: val_loss did not improve from 0.00858
90/90 - 0s - loss: 7.7845e-04 - accuracy: 0.0000e+00 - val_loss: 0.0211 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 475ms/epoch - 5ms/step
Epoch 00064: early stopping
SMA Prediction vs Close: 53.73% Accuracy Prediction vs Prediction: 47.01% Accuracy MSE: 39.87524298067742 RMSE: 6.3146847095225125 MAPE: 5.088204561858354 EMA Prediction vs Close: 52.61% Accuracy Prediction vs Prediction: 43.66% Accuracy MSE: 73.6938994303845 RMSE: 8.584515095821342 MAPE: 7.207683534137507 WMA Prediction vs Close: 52.99% Accuracy Prediction vs Prediction: 45.52% Accuracy MSE: 76.90233333404232 RMSE: 8.76939754681257 MAPE: 7.121360770950225 DEMA Prediction vs Close: 52.24% Accuracy Prediction vs Prediction: 46.27% Accuracy MSE: 131.0602137141292 RMSE: 11.448153288374908 MAPE: 10.329401784343453 KAMA Prediction vs Close: 53.36% Accuracy Prediction vs Prediction: 47.76% Accuracy MSE: 42.128331463904196 RMSE: 6.49063413418937 MAPE: 5.225311201416733 MIDPOINT Prediction vs Close: 52.24% Accuracy Prediction vs Prediction: 45.15% Accuracy MSE: 102.72361049298229 RMSE: 10.135265684380567 MAPE: 8.372939983805384 T3 Prediction vs Close: 55.22% Accuracy Prediction vs Prediction: 45.15% Accuracy MSE: 106.78678832331099 RMSE: 10.333769318274479 MAPE: 8.334873887974807 TEMA Prediction vs Close: 50.75% Accuracy Prediction vs Prediction: 48.88% Accuracy MSE: 78.66857412996737 RMSE: 8.86953066007257 MAPE: 7.566655356961055 Runtime: mins: 11.6630284996
Architecture Used¶
In [98]:
from google.colab import files
import cv2
uploaded = files.upload()
In [99]:
img = cv2.imread('Experiment2.png')
plt.figure(figsize=(20,10))
plt.axis("off")
plt.title('LSTM Architecture '+imgfile,fontsize=18)
plt.imshow(img)
Out[99]:
<matplotlib.image.AxesImage at 0x7f75c042c4d0>
Model Plots¶
In [99]:
import json
with open('simulation2_data.json') as json_file:
simulation2 = json.load(json_file)
fileimg = 'Experiment2'
In [100]:
for i in range(len(list(simulation2.keys()))):
SIM = list(simulation2.keys())[i]
plot_train(simulation2,SIM)
plot_test(simulation2,SIM)
----- Train RMSE for SMA ----- 8.889283880470046 ----- Train_MSE_LSTM for SMA ----- 79.0193679075846 ----- Train MAE LSTM for SMA ----- 7.8115937103860835
----- Test RMSE for SMA----- 6.3146847095225125 ----- Test_MSE_LSTM for SMA----- 39.87524298067742 ----- Test_MAE_LSTM for SMA----- 5.088204561858354
----- Train RMSE for EMA ----- 10.150148453019417 ----- Train_MSE_LSTM for EMA ----- 103.02551361833245 ----- Train MAE LSTM for EMA ----- 8.952000885856286
----- Test RMSE for EMA----- 8.584515095821342 ----- Test_MSE_LSTM for EMA----- 73.6938994303845 ----- Test_MAE_LSTM for EMA----- 7.207683534137507
----- Train RMSE for WMA ----- 10.467180278295531 ----- Train_MSE_LSTM for WMA ----- 109.56186297833894 ----- Train MAE LSTM for WMA ----- 9.323534322736583
----- Test RMSE for WMA----- 8.76939754681257 ----- Test_MSE_LSTM for WMA----- 76.90233333404232 ----- Test_MAE_LSTM for WMA----- 7.121360770950225
----- Train RMSE for DEMA ----- 12.075408120538793 ----- Train_MSE_LSTM for DEMA ----- 145.81548127757424 ----- Train MAE LSTM for DEMA ----- 10.813702843841257
----- Test RMSE for DEMA----- 11.448153288374908 ----- Test_MSE_LSTM for DEMA----- 131.0602137141292 ----- Test_MAE_LSTM for DEMA----- 10.329401784343453
----- Train RMSE for KAMA ----- 10.56352496627593 ----- Train_MSE_LSTM for KAMA ----- 111.58805971313488 ----- Train MAE LSTM for KAMA ----- 9.51725184300564
----- Test RMSE for KAMA----- 6.49063413418937 ----- Test_MSE_LSTM for KAMA----- 42.128331463904196 ----- Test_MAE_LSTM for KAMA----- 5.225311201416733
----- Train RMSE for MIDPOINT ----- 9.508048912895521 ----- Train_MSE_LSTM for MIDPOINT ----- 90.4029941300137 ----- Train MAE LSTM for MIDPOINT ----- 8.394449433621876
----- Test RMSE for MIDPOINT----- 10.135265684380567 ----- Test_MSE_LSTM for MIDPOINT----- 102.72361049298229 ----- Test_MAE_LSTM for MIDPOINT----- 8.372939983805384
----- Train RMSE for T3 ----- 12.069596897007075 ----- Train_MSE_LSTM for T3 ----- 145.67516925624284 ----- Train MAE LSTM for T3 ----- 10.870868175127285
----- Test RMSE for T3----- 10.333769318274479 ----- Test_MSE_LSTM for T3----- 106.78678832331099 ----- Test_MAE_LSTM for T3----- 8.334873887974807
----- Train RMSE for TEMA ----- 7.440796513329569 ----- Train_MSE_LSTM for TEMA ----- 55.36545275277747 ----- Train MAE LSTM for TEMA ----- 5.170842815856772
----- Test RMSE for TEMA----- 8.86953066007257 ----- Test_MSE_LSTM for TEMA----- 78.66857412996737 ----- Test_MAE_LSTM for TEMA----- 7.566655356961055
Univariate Arima Multistep MutiVariate LSTM Hybrid Model Experiment 3¶
In [101]:
def get_lstm(data,original_data, train_len, test_len,img_file,ma ,lstm_len=3):
# prepare train and test data
X_value = pd.DataFrame(data.iloc[:, :])
y_value = pd.DataFrame(data.iloc[:, 3])
X_scaler = MinMaxScaler(feature_range=(-1, 1))
y_scaler = MinMaxScaler(feature_range=(-1, 1))
X_scaler.fit(X_value)
y_scaler.fit(y_value)
# Get data and check shape
X, y, yc = get_X_y(X_scale_dataset, y_scale_dataset)#X will be of shape 224 X 3 X 21 (each 3 X 21 array will be 3 days' worth of data). yc will have the corresponding closing price value
# pdb.set_trace()
X_train, X_test, = split_train_test(X)
y_train, y_test, = split_train_test(y)
# yc_train, yc_test, = split_train_test(original_data)
index_train, index_test, = predict_index(dataset_final, X_train, n_steps_in, n_steps_out)
det = 20
input_dim = X_train.shape[1]#3
feature_size = X_train.shape[2]#24
output_dim = y_train.shape[1]#1
# # Option 1
# # Set up & fit LSTM RNN
# model = Sequential()
# model.add(LSTM(256, activation='relu', kernel_initializer='he_normal', input_shape=(input_dim, feature_size)))
# model.add(Dense(units=64,activation='relu'))
# model.add(Dropout(0.5))
# model.add(Dense(units=output_dim))
# model.compile(optimizer=Adam(learning_rate = 0.001), loss='mse')
# ## Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# # option 2
# model = Sequential()
# model.add(Bidirectional(LSTM(units= 128), input_shape=(input_dim, feature_size)))
# model.add(Dense(64))
# model.add(Dense(units=output_dim))
# model.compile(optimizer=Adam(lr = 0.001), loss='mean_squared_error', metrics=['accuracy'])
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# Option 3
# define custom activation
#
class Double_Tanh(Activation):
def __init__(self, activation, **kwargs):
super(Double_Tanh, self).__init__(activation, **kwargs)
self.__name__ = 'double_tanh'
def double_tanh(x):
return (K.tanh(x) * 2)
get_custom_objects().update({'double_tanh':Double_Tanh(double_tanh)})
# Model Generation
model = Sequential()
#check https://machinelearningmastery.com/use-weight-regularization-lstm-networks-time-series-forecasting/
model.add(LSTM(25, input_shape=(input_dim, feature_size), dropout=0.2, kernel_regularizer=l1_l2(0.00,0.00), bias_regularizer=l1_l2(0.00,0.00)))
model.add(Dense(1))
model.add(Activation(double_tanh))
model.compile(loss='mean_squared_error', optimizer='adam', metrics=['mse', 'mae'])
# Common code
callbacks = [
EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
ModelCheckpoint('LSTM3.h5', verbose=1, save_best_only=True, save_weights_only=True)]
fname1 = img_file+'.png'
tensorflow.keras.utils.plot_model(
model, to_file=fname1, show_shapes=True, show_dtype=False,
show_layer_names=True, expand_nested=False, dpi=96,
layer_range=None, show_layer_activations=False
)
history = model.fit(X_train, y_train, epochs=500, batch_size=int( optimized_period[ma]), verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# plot loss
fname2 = img_file+'-'+ma
plt.title(img_file+'-'+ma+' Loss')
plt.xlabel("Epochs")
plt.ylabel("Loss")
pyplot.plot(history.history['loss'], label='train')
pyplot.plot(history.history['val_loss'], label='validation')
pyplot.legend()
pyplot.savefig(fname2+'.png',dpi='figure')
pyplot.show()
# Option 4
# Set up & fit LSTM RNN
# model = Sequential()
# model.add(LSTM(units=lstm_len, return_sequences=True, input_shape=(x_train.shape[1], 1)))
# model.add(LSTM(units=int(lstm_len/2)))
# model.add(Dense(1, activation='sigmoid'))
# model.compile(loss='mean_squared_error', optimizer='adam')
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# Generate predictions
predictiontr = model.predict(X_train, verbose=0)
predictiontr = y_scaler.inverse_transform(predictiontr).tolist()
outputtr = []
for i in range(len(predictiontr)):
outputtr.extend(predictiontr[i])
predictiontr = outputtr
# Generate error data
## replace with yc , xtest generated by new multistep method
mse_tr = mean_squared_error(y_train, predictiontr)
rmse_tr = mse_tr ** 0.5
# mape = mean_absolute_percentage_error(X_test, pd.Series(predictiontr))
mae_tr = mean_absolute_error(y_train, pd.Series(predictiontr))
# Original_tr = pd.Series(yc_train)
Original_tr = y_scaler.inverse_transform(y_train).flatten().tolist()
predictionte = model.predict(X_test, verbose=0)
predictionte = (y_scaler.inverse_transform(predictionte)-det).tolist()
outputte = []
for i in range(len(predictionte)):
outputte.extend(predictionte[i])
predictionte = outputte
# Generate error data
mse_te = mean_squared_error(y_test, predictionte)
rmse_te = mse_te ** 0.5
# mape = mean_absolute_percentage_error(X_test, pd.Series(predictionte))
mae_te = mean_absolute_error(y_test, pd.Series(predictionte))
# Original_te = pd.Series(yc_test)
Original_te = y_scaler.inverse_transform(y_test).flatten().tolist()
return Original_tr, predictiontr, mse_tr, rmse_tr,mae_tr,Original_te,predictionte, mse_te,rmse_te,mae_te
In [102]:
if __name__ == '__main__':
start_time = timeit.default_timer()
simulation3 = {}
imgfile = 'Experiment3'
for ma in optimized_period:
print(ma)
print(functions[ma])
print ( int( optimized_period[ma]))
# if ma == 'SMA':
low_vol = df.apply(lambda c: functions[ma](c, timeperiod = int( optimized_period[ma])))
low_vol = low_vol.fillna(0)
low_vol_data = df['close']
high_vol = pd.DataFrame()
df2 = df.copy()
for i in df2.columns:
if i in low_vol.columns:
high_vol[i] = df2[i].subtract(low_vol[i], fill_value=0)
high_vol_data = df['close']
## *****************************************************
# Generate ARIMA and LSTM predictions
print('\nWorking on ' + ma + ' predictions')
try:
print('parameters used : ', train_len, test_len)
low_vol_Original, low_vol_prediction, low_vol_mse, low_vol_rmse,low_vol_mae = get_arima(low_vol,low_vol_data, train_len, test_len)
except:
print('ARIMA error, skipping to next MA type')
continue
Original_tr, predictiontr, mse_tr, rmse_tr,mae_tr,high_vol_Original, high_vol_prediction, high_vol_mse, high_vol_rmse,high_vol_mae, = get_lstm(high_vol,high_vol_data, train_len, test_len,imgfile,ma)
final_prediction_tr = df['close'].head(train_len).values + pd.Series(predictiontr) # ignoring first 3 steps
mse_ftr = mean_squared_error(df['close'].head(train_len).values,final_prediction_tr.values)
rmse_ftr = mse_ftr ** 0.5
mape_ftr = mean_absolute_percentage_error(df['close'].head(train_len).reset_index(drop=True), final_prediction_tr)
mae_ftr = mean_absolute_error(df['close'].head(train_len).reset_index(drop=True), final_prediction_tr)
final_prediction = pd.Series(low_vol_prediction[3:]) + pd.Series(high_vol_prediction)
mse = mean_squared_error(df['close'].tail(test_len).values,final_prediction.values)
rmse = mse ** 0.5
mape = mean_absolute_percentage_error(df['close'].tail(test_len).reset_index(drop=True), final_prediction)
mae = mean_absolute_error(df['close'].tail(test_len).reset_index(drop=True), final_prediction)
# Generate prediction accuracy
actual = df['close'].tail(test_len).values
result_1 = []
result_2 = []
for i in range(1, len(final_prediction)):
# Compare prediction to previous close price
if final_prediction[i] > actual[i-1] and actual[i] > actual[i-1]:
result_1.append(1)
elif final_prediction[i] < actual[i-1] and actual[i] < actual[i-1]:
result_1.append(1)
else:
result_1.append(0)
# Compare prediction to previous prediction
if final_prediction[i] > final_prediction[i-1] and actual[i] > actual[i-1]:
result_2.append(1)
elif final_prediction[i] < final_prediction[i-1] and actual[i] < actual[i-1]:
result_2.append(1)
else:
result_2.append(0)
accuracy_1 = np.mean(result_1)
accuracy_2 = np.mean(result_2)
simulation3[ma] = {'low_vol': {'original':list(low_vol_Original), 'prediction': list(low_vol_prediction) , 'mse': low_vol_mse,
'rmse': low_vol_rmse, 'mae' : low_vol_mae},
'high_vol': {'original':list(high_vol_Original),'prediction': list(high_vol_prediction), 'mse': high_vol_mse,
'rmse': high_vol_rmse, 'mae' : high_vol_mae},
'final_tr': {'original':df['close'].head(train_len).tolist(),'prediction': final_prediction_tr.values.tolist(), 'mse': mse_ftr,
'rmse': rmse_ftr, 'mae' : mae_ftr},
'final': {'original': df['close'].tail(test_len).tolist(), 'prediction': final_prediction.values.tolist(), 'mse': mse,
'rmse': rmse, 'mae': mae },
'accuracy': {'prediction vs close': accuracy_1, 'prediction vs prediction': accuracy_2}}
# save simulation data here as checkpoint
with open('simulation3_data.json', 'w') as fp:
json.dump(simulation3, fp)
for ma in simulation3.keys():
print('\n' + ma)
print('Prediction vs Close:\t\t' + str(round(100*simulation3[ma]['accuracy']['prediction vs close'], 2))
+ '% Accuracy')
print('Prediction vs Prediction:\t' + str(round(100*simulation3[ma]['accuracy']['prediction vs prediction'], 2))
+ '% Accuracy')
print('MSE:\t', simulation3[ma]['final']['mse'],
'\nRMSE:\t', simulation3[ma]['final']['rmse'],
'\nMAPE:\t', simulation3[ma]['final']['mae'])#,
# '\nMAPE:\t', simulation[ma]['final']['mape'])
# else:
# break
elapsed = timeit.default_timer() - start_time
print('Runtime: mins:',elapsed/60)
SMA
SMA([input_arrays], [timeperiod=30])
Simple Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
17
Working on SMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.47 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4157.020, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3687.148, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.21 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3458.651, Time=0.07 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3322.133, Time=0.09 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=0.74 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.81 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3324.133, Time=0.23 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 2.688 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1657.067
Date: Sun, 12 Dec 2021 AIC 3322.133
Time: 14:16:48 BIC 3340.897
Sample: 0 HQIC 3329.339
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1966 0.003 -387.226 0.000 -1.203 -1.191
ar.L2 -0.8952 0.006 -138.692 0.000 -0.908 -0.883
ar.L3 -0.3968 0.006 -68.284 0.000 -0.408 -0.385
sigma2 3.5858 0.017 214.535 0.000 3.553 3.619
===================================================================================
Ljung-Box (L1) (Q): 14.47 Jarque-Bera (JB): 2428881.42
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.90
Prob(H) (two-sided): 0.00 Kurtosis: 271.99
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.31248, saving model to LSTM3.h5
48/48 - 3s - loss: 0.1235 - mse: 0.1235 - mae: 0.2748 - val_loss: 0.3125 - val_mse: 0.3125 - val_mae: 0.5302 - lr: 0.0010 - 3s/epoch - 53ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.31248 to 0.04955, saving model to LSTM3.h5
48/48 - 0s - loss: 0.0349 - mse: 0.0349 - mae: 0.1484 - val_loss: 0.0495 - val_mse: 0.0495 - val_mae: 0.1877 - lr: 0.0010 - 240ms/epoch - 5ms/step
Epoch 3/500
Epoch 00003: val_loss improved from 0.04955 to 0.01856, saving model to LSTM3.h5
48/48 - 0s - loss: 0.0240 - mse: 0.0240 - mae: 0.1235 - val_loss: 0.0186 - val_mse: 0.0186 - val_mae: 0.1079 - lr: 0.0010 - 264ms/epoch - 6ms/step
Epoch 4/500
Epoch 00004: val_loss improved from 0.01856 to 0.01553, saving model to LSTM3.h5
48/48 - 0s - loss: 0.0184 - mse: 0.0184 - mae: 0.1091 - val_loss: 0.0155 - val_mse: 0.0155 - val_mae: 0.1005 - lr: 0.0010 - 290ms/epoch - 6ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0144 - mse: 0.0144 - mae: 0.0966 - val_loss: 0.0197 - val_mse: 0.0197 - val_mae: 0.1114 - lr: 0.0010 - 231ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0131 - mse: 0.0131 - mae: 0.0908 - val_loss: 0.0208 - val_mse: 0.0208 - val_mae: 0.1150 - lr: 0.0010 - 217ms/epoch - 5ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0143 - mse: 0.0143 - mae: 0.0940 - val_loss: 0.0247 - val_mse: 0.0247 - val_mae: 0.1262 - lr: 0.0010 - 205ms/epoch - 4ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0134 - mse: 0.0134 - mae: 0.0943 - val_loss: 0.0278 - val_mse: 0.0278 - val_mae: 0.1345 - lr: 0.0010 - 203ms/epoch - 4ms/step
Epoch 9/500
Epoch 00009: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00009: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0142 - mse: 0.0142 - mae: 0.0954 - val_loss: 0.0357 - val_mse: 0.0357 - val_mae: 0.1546 - lr: 0.0010 - 209ms/epoch - 4ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0163 - mse: 0.0163 - mae: 0.0991 - val_loss: 0.0292 - val_mse: 0.0292 - val_mae: 0.1386 - lr: 1.0000e-04 - 229ms/epoch - 5ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0139 - mse: 0.0139 - mae: 0.0954 - val_loss: 0.0304 - val_mse: 0.0304 - val_mae: 0.1413 - lr: 1.0000e-04 - 197ms/epoch - 4ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0110 - mse: 0.0110 - mae: 0.0846 - val_loss: 0.0307 - val_mse: 0.0307 - val_mae: 0.1419 - lr: 1.0000e-04 - 237ms/epoch - 5ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0106 - mse: 0.0106 - mae: 0.0830 - val_loss: 0.0325 - val_mse: 0.0325 - val_mae: 0.1464 - lr: 1.0000e-04 - 268ms/epoch - 6ms/step
Epoch 14/500
Epoch 00014: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00014: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0090 - mse: 0.0090 - mae: 0.0769 - val_loss: 0.0330 - val_mse: 0.0330 - val_mae: 0.1478 - lr: 1.0000e-04 - 222ms/epoch - 5ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0077 - mse: 0.0077 - mae: 0.0708 - val_loss: 0.0334 - val_mse: 0.0334 - val_mae: 0.1488 - lr: 1.0000e-05 - 217ms/epoch - 5ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0082 - mse: 0.0082 - mae: 0.0715 - val_loss: 0.0336 - val_mse: 0.0336 - val_mae: 0.1493 - lr: 1.0000e-05 - 195ms/epoch - 4ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0082 - mse: 0.0082 - mae: 0.0728 - val_loss: 0.0339 - val_mse: 0.0339 - val_mae: 0.1502 - lr: 1.0000e-05 - 235ms/epoch - 5ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0076 - mse: 0.0076 - mae: 0.0698 - val_loss: 0.0344 - val_mse: 0.0344 - val_mae: 0.1513 - lr: 1.0000e-05 - 246ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00019: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0078 - mse: 0.0078 - mae: 0.0703 - val_loss: 0.0344 - val_mse: 0.0344 - val_mae: 0.1514 - lr: 1.0000e-05 - 222ms/epoch - 5ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0077 - mse: 0.0077 - mae: 0.0698 - val_loss: 0.0347 - val_mse: 0.0347 - val_mae: 0.1522 - lr: 1.0000e-05 - 178ms/epoch - 4ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0077 - mse: 0.0077 - mae: 0.0702 - val_loss: 0.0347 - val_mse: 0.0347 - val_mae: 0.1523 - lr: 1.0000e-05 - 218ms/epoch - 5ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0082 - mse: 0.0082 - mae: 0.0712 - val_loss: 0.0349 - val_mse: 0.0349 - val_mae: 0.1527 - lr: 1.0000e-05 - 236ms/epoch - 5ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0070 - mse: 0.0070 - mae: 0.0670 - val_loss: 0.0350 - val_mse: 0.0350 - val_mae: 0.1530 - lr: 1.0000e-05 - 230ms/epoch - 5ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0070 - mse: 0.0070 - mae: 0.0667 - val_loss: 0.0352 - val_mse: 0.0352 - val_mae: 0.1534 - lr: 1.0000e-05 - 219ms/epoch - 5ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0067 - mse: 0.0067 - mae: 0.0643 - val_loss: 0.0355 - val_mse: 0.0355 - val_mae: 0.1542 - lr: 1.0000e-05 - 225ms/epoch - 5ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0074 - mse: 0.0074 - mae: 0.0687 - val_loss: 0.0357 - val_mse: 0.0357 - val_mae: 0.1548 - lr: 1.0000e-05 - 207ms/epoch - 4ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0070 - mse: 0.0070 - mae: 0.0657 - val_loss: 0.0358 - val_mse: 0.0358 - val_mae: 0.1549 - lr: 1.0000e-05 - 235ms/epoch - 5ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0071 - mse: 0.0071 - mae: 0.0669 - val_loss: 0.0361 - val_mse: 0.0361 - val_mae: 0.1558 - lr: 1.0000e-05 - 236ms/epoch - 5ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0073 - mse: 0.0073 - mae: 0.0674 - val_loss: 0.0361 - val_mse: 0.0361 - val_mae: 0.1558 - lr: 1.0000e-05 - 228ms/epoch - 5ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0074 - mse: 0.0074 - mae: 0.0677 - val_loss: 0.0363 - val_mse: 0.0363 - val_mae: 0.1562 - lr: 1.0000e-05 - 217ms/epoch - 5ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0067 - mse: 0.0067 - mae: 0.0648 - val_loss: 0.0366 - val_mse: 0.0366 - val_mae: 0.1570 - lr: 1.0000e-05 - 239ms/epoch - 5ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0075 - mse: 0.0075 - mae: 0.0687 - val_loss: 0.0366 - val_mse: 0.0366 - val_mae: 0.1568 - lr: 1.0000e-05 - 230ms/epoch - 5ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0068 - mse: 0.0068 - mae: 0.0655 - val_loss: 0.0368 - val_mse: 0.0368 - val_mae: 0.1573 - lr: 1.0000e-05 - 239ms/epoch - 5ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0651 - val_loss: 0.0371 - val_mse: 0.0371 - val_mae: 0.1581 - lr: 1.0000e-05 - 213ms/epoch - 4ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0070 - mse: 0.0070 - mae: 0.0653 - val_loss: 0.0372 - val_mse: 0.0372 - val_mae: 0.1583 - lr: 1.0000e-05 - 231ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0073 - mse: 0.0073 - mae: 0.0660 - val_loss: 0.0374 - val_mse: 0.0374 - val_mae: 0.1590 - lr: 1.0000e-05 - 217ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0652 - val_loss: 0.0380 - val_mse: 0.0380 - val_mae: 0.1602 - lr: 1.0000e-05 - 243ms/epoch - 5ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0638 - val_loss: 0.0383 - val_mse: 0.0383 - val_mae: 0.1610 - lr: 1.0000e-05 - 229ms/epoch - 5ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0641 - val_loss: 0.0385 - val_mse: 0.0385 - val_mae: 0.1614 - lr: 1.0000e-05 - 216ms/epoch - 5ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0634 - val_loss: 0.0386 - val_mse: 0.0386 - val_mae: 0.1617 - lr: 1.0000e-05 - 201ms/epoch - 4ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0067 - mse: 0.0067 - mae: 0.0646 - val_loss: 0.0384 - val_mse: 0.0384 - val_mae: 0.1613 - lr: 1.0000e-05 - 223ms/epoch - 5ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0068 - mse: 0.0068 - mae: 0.0633 - val_loss: 0.0384 - val_mse: 0.0384 - val_mae: 0.1614 - lr: 1.0000e-05 - 231ms/epoch - 5ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0618 - val_loss: 0.0383 - val_mse: 0.0383 - val_mae: 0.1611 - lr: 1.0000e-05 - 220ms/epoch - 5ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0067 - mse: 0.0067 - mae: 0.0641 - val_loss: 0.0381 - val_mse: 0.0381 - val_mae: 0.1605 - lr: 1.0000e-05 - 230ms/epoch - 5ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0058 - mse: 0.0058 - mae: 0.0600 - val_loss: 0.0383 - val_mse: 0.0383 - val_mae: 0.1612 - lr: 1.0000e-05 - 200ms/epoch - 4ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0647 - val_loss: 0.0388 - val_mse: 0.0388 - val_mae: 0.1623 - lr: 1.0000e-05 - 195ms/epoch - 4ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0636 - val_loss: 0.0390 - val_mse: 0.0390 - val_mae: 0.1627 - lr: 1.0000e-05 - 223ms/epoch - 5ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0638 - val_loss: 0.0391 - val_mse: 0.0391 - val_mae: 0.1629 - lr: 1.0000e-05 - 237ms/epoch - 5ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0617 - val_loss: 0.0392 - val_mse: 0.0392 - val_mae: 0.1632 - lr: 1.0000e-05 - 214ms/epoch - 4ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0624 - val_loss: 0.0394 - val_mse: 0.0394 - val_mae: 0.1638 - lr: 1.0000e-05 - 225ms/epoch - 5ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0584 - val_loss: 0.0395 - val_mse: 0.0395 - val_mae: 0.1641 - lr: 1.0000e-05 - 235ms/epoch - 5ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0063 - mse: 0.0063 - mae: 0.0621 - val_loss: 0.0396 - val_mse: 0.0396 - val_mae: 0.1643 - lr: 1.0000e-05 - 231ms/epoch - 5ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0624 - val_loss: 0.0395 - val_mse: 0.0395 - val_mae: 0.1639 - lr: 1.0000e-05 - 238ms/epoch - 5ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.01553
48/48 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0613 - val_loss: 0.0392 - val_mse: 0.0392 - val_mae: 0.1632 - lr: 1.0000e-05 - 227ms/epoch - 5ms/step
Epoch 00054: early stopping
SMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 50.75% Accuracy
MSE: 39.71707698339657
RMSE: 6.302148600548591
MAPE: 5.205453851116191
EMA
EMA([input_arrays], [timeperiod=30])
Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
51
Working on EMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.42 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4231.556, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3761.238, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.30 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3532.227, Time=0.09 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3394.496, Time=0.10 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=0.87 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.67 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3396.496, Time=0.23 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 2.742 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1693.248
Date: Sun, 12 Dec 2021 AIC 3394.496
Time: 14:18:13 BIC 3413.260
Sample: 0 HQIC 3401.702
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1982 0.003 -389.569 0.000 -1.204 -1.192
ar.L2 -0.8976 0.006 -139.811 0.000 -0.910 -0.885
ar.L3 -0.3984 0.006 -68.662 0.000 -0.410 -0.387
sigma2 3.9230 0.018 215.372 0.000 3.887 3.959
===================================================================================
Ljung-Box (L1) (Q): 14.54 Jarque-Bera (JB): 2462173.05
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.90
Prob(H) (two-sided): 0.00 Kurtosis: 273.82
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.40077, saving model to LSTM3.h5
16/16 - 2s - loss: 0.3798 - mse: 0.3798 - mae: 0.4758 - val_loss: 0.4008 - val_mse: 0.4008 - val_mae: 0.6029 - lr: 0.0010 - 2s/epoch - 144ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.40077 to 0.34145, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0757 - mse: 0.0757 - mae: 0.2395 - val_loss: 0.3414 - val_mse: 0.3414 - val_mae: 0.5546 - lr: 0.0010 - 98ms/epoch - 6ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.34145
16/16 - 0s - loss: 0.0330 - mse: 0.0330 - mae: 0.1493 - val_loss: 0.3493 - val_mse: 0.3493 - val_mae: 0.5631 - lr: 0.0010 - 81ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss improved from 0.34145 to 0.27846, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0299 - mse: 0.0299 - mae: 0.1395 - val_loss: 0.2785 - val_mse: 0.2785 - val_mae: 0.4979 - lr: 0.0010 - 87ms/epoch - 5ms/step
Epoch 5/500
Epoch 00005: val_loss improved from 0.27846 to 0.24516, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0210 - mse: 0.0210 - mae: 0.1153 - val_loss: 0.2452 - val_mse: 0.2452 - val_mae: 0.4639 - lr: 0.0010 - 87ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: val_loss improved from 0.24516 to 0.22130, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0196 - mse: 0.0196 - mae: 0.1145 - val_loss: 0.2213 - val_mse: 0.2213 - val_mae: 0.4379 - lr: 0.0010 - 102ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: val_loss improved from 0.22130 to 0.20755, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0166 - mse: 0.0166 - mae: 0.1031 - val_loss: 0.2076 - val_mse: 0.2076 - val_mae: 0.4221 - lr: 0.0010 - 128ms/epoch - 8ms/step
Epoch 8/500
Epoch 00008: val_loss improved from 0.20755 to 0.18812, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0160 - mse: 0.0160 - mae: 0.0998 - val_loss: 0.1881 - val_mse: 0.1881 - val_mae: 0.3987 - lr: 0.0010 - 118ms/epoch - 7ms/step
Epoch 9/500
Epoch 00009: val_loss improved from 0.18812 to 0.18203, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0133 - mse: 0.0133 - mae: 0.0923 - val_loss: 0.1820 - val_mse: 0.1820 - val_mae: 0.3915 - lr: 0.0010 - 106ms/epoch - 7ms/step
Epoch 10/500
Epoch 00010: val_loss improved from 0.18203 to 0.16938, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0128 - mse: 0.0128 - mae: 0.0896 - val_loss: 0.1694 - val_mse: 0.1694 - val_mae: 0.3756 - lr: 0.0010 - 101ms/epoch - 6ms/step
Epoch 11/500
Epoch 00011: val_loss improved from 0.16938 to 0.16630, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0113 - mse: 0.0113 - mae: 0.0847 - val_loss: 0.1663 - val_mse: 0.1663 - val_mae: 0.3715 - lr: 0.0010 - 102ms/epoch - 6ms/step
Epoch 12/500
Epoch 00012: val_loss improved from 0.16630 to 0.15845, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0111 - mse: 0.0111 - mae: 0.0834 - val_loss: 0.1585 - val_mse: 0.1585 - val_mae: 0.3605 - lr: 0.0010 - 97ms/epoch - 6ms/step
Epoch 13/500
Epoch 00013: val_loss improved from 0.15845 to 0.15831, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0103 - mse: 0.0103 - mae: 0.0817 - val_loss: 0.1583 - val_mse: 0.1583 - val_mae: 0.3603 - lr: 0.0010 - 96ms/epoch - 6ms/step
Epoch 14/500
Epoch 00014: val_loss improved from 0.15831 to 0.15122, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0107 - mse: 0.0107 - mae: 0.0824 - val_loss: 0.1512 - val_mse: 0.1512 - val_mae: 0.3505 - lr: 0.0010 - 90ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.15122
16/16 - 0s - loss: 0.0091 - mse: 0.0091 - mae: 0.0758 - val_loss: 0.1529 - val_mse: 0.1529 - val_mae: 0.3525 - lr: 0.0010 - 81ms/epoch - 5ms/step
Epoch 16/500
Epoch 00016: val_loss improved from 0.15122 to 0.14976, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0092 - mse: 0.0092 - mae: 0.0745 - val_loss: 0.1498 - val_mse: 0.1498 - val_mae: 0.3479 - lr: 0.0010 - 106ms/epoch - 7ms/step
Epoch 17/500
Epoch 00017: val_loss improved from 0.14976 to 0.14433, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0086 - mse: 0.0086 - mae: 0.0739 - val_loss: 0.1443 - val_mse: 0.1443 - val_mae: 0.3401 - lr: 0.0010 - 132ms/epoch - 8ms/step
Epoch 18/500
Epoch 00018: val_loss improved from 0.14433 to 0.13994, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0085 - mse: 0.0085 - mae: 0.0729 - val_loss: 0.1399 - val_mse: 0.1399 - val_mae: 0.3338 - lr: 0.0010 - 111ms/epoch - 7ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.13994
16/16 - 0s - loss: 0.0074 - mse: 0.0074 - mae: 0.0678 - val_loss: 0.1414 - val_mse: 0.1414 - val_mae: 0.3357 - lr: 0.0010 - 95ms/epoch - 6ms/step
Epoch 20/500
Epoch 00020: val_loss improved from 0.13994 to 0.13241, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0077 - mse: 0.0077 - mae: 0.0698 - val_loss: 0.1324 - val_mse: 0.1324 - val_mae: 0.3227 - lr: 0.0010 - 115ms/epoch - 7ms/step
Epoch 21/500
Epoch 00021: val_loss improved from 0.13241 to 0.13227, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0069 - mse: 0.0069 - mae: 0.0659 - val_loss: 0.1323 - val_mse: 0.1323 - val_mae: 0.3228 - lr: 0.0010 - 120ms/epoch - 7ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.13227
16/16 - 0s - loss: 0.0068 - mse: 0.0068 - mae: 0.0650 - val_loss: 0.1329 - val_mse: 0.1329 - val_mae: 0.3239 - lr: 0.0010 - 92ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: val_loss improved from 0.13227 to 0.12353, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0642 - val_loss: 0.1235 - val_mse: 0.1235 - val_mae: 0.3100 - lr: 0.0010 - 101ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss improved from 0.12353 to 0.12175, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0065 - mse: 0.0065 - mae: 0.0632 - val_loss: 0.1217 - val_mse: 0.1217 - val_mae: 0.3071 - lr: 0.0010 - 94ms/epoch - 6ms/step
Epoch 25/500
Epoch 00025: val_loss improved from 0.12175 to 0.11701, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0063 - mse: 0.0063 - mae: 0.0617 - val_loss: 0.1170 - val_mse: 0.1170 - val_mae: 0.2999 - lr: 0.0010 - 104ms/epoch - 7ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.11701
16/16 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0580 - val_loss: 0.1190 - val_mse: 0.1190 - val_mae: 0.3033 - lr: 0.0010 - 90ms/epoch - 6ms/step
Epoch 27/500
Epoch 00027: val_loss improved from 0.11701 to 0.10937, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0605 - val_loss: 0.1094 - val_mse: 0.1094 - val_mae: 0.2885 - lr: 0.0010 - 91ms/epoch - 6ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.10937
16/16 - 0s - loss: 0.0058 - mse: 0.0058 - mae: 0.0597 - val_loss: 0.1104 - val_mse: 0.1104 - val_mae: 0.2904 - lr: 0.0010 - 86ms/epoch - 5ms/step
Epoch 29/500
Epoch 00029: val_loss improved from 0.10937 to 0.10331, saving model to LSTM3.h5
16/16 - 0s - loss: 0.0053 - mse: 0.0053 - mae: 0.0577 - val_loss: 0.1033 - val_mse: 0.1033 - val_mae: 0.2785 - lr: 0.0010 - 117ms/epoch - 7ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0053 - mse: 0.0053 - mae: 0.0573 - val_loss: 0.1096 - val_mse: 0.1096 - val_mae: 0.2888 - lr: 0.0010 - 83ms/epoch - 5ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0052 - mse: 0.0052 - mae: 0.0562 - val_loss: 0.1106 - val_mse: 0.1106 - val_mae: 0.2908 - lr: 0.0010 - 84ms/epoch - 5ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0050 - mse: 0.0050 - mae: 0.0550 - val_loss: 0.1144 - val_mse: 0.1144 - val_mae: 0.2967 - lr: 0.0010 - 78ms/epoch - 5ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0522 - val_loss: 0.1160 - val_mse: 0.1160 - val_mae: 0.2994 - lr: 0.0010 - 79ms/epoch - 5ms/step
Epoch 34/500
Epoch 00034: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00034: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0536 - val_loss: 0.1075 - val_mse: 0.1075 - val_mae: 0.2858 - lr: 0.0010 - 86ms/epoch - 5ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0052 - mse: 0.0052 - mae: 0.0560 - val_loss: 0.1077 - val_mse: 0.1077 - val_mae: 0.2862 - lr: 1.0000e-04 - 77ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0043 - mse: 0.0043 - mae: 0.0523 - val_loss: 0.1081 - val_mse: 0.1081 - val_mae: 0.2870 - lr: 1.0000e-04 - 82ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0043 - mse: 0.0043 - mae: 0.0499 - val_loss: 0.1082 - val_mse: 0.1082 - val_mae: 0.2871 - lr: 1.0000e-04 - 104ms/epoch - 7ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0509 - val_loss: 0.1078 - val_mse: 0.1078 - val_mae: 0.2866 - lr: 1.0000e-04 - 91ms/epoch - 6ms/step
Epoch 39/500
Epoch 00039: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00039: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0502 - val_loss: 0.1084 - val_mse: 0.1084 - val_mae: 0.2875 - lr: 1.0000e-04 - 112ms/epoch - 7ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0517 - val_loss: 0.1083 - val_mse: 0.1083 - val_mae: 0.2874 - lr: 1.0000e-05 - 104ms/epoch - 6ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0527 - val_loss: 0.1082 - val_mse: 0.1082 - val_mae: 0.2873 - lr: 1.0000e-05 - 97ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0529 - val_loss: 0.1082 - val_mse: 0.1082 - val_mae: 0.2872 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0522 - val_loss: 0.1082 - val_mse: 0.1082 - val_mae: 0.2871 - lr: 1.0000e-05 - 96ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00044: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0045 - mse: 0.0045 - mae: 0.0527 - val_loss: 0.1081 - val_mse: 0.1081 - val_mae: 0.2871 - lr: 1.0000e-05 - 84ms/epoch - 5ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0535 - val_loss: 0.1081 - val_mse: 0.1081 - val_mae: 0.2871 - lr: 1.0000e-05 - 93ms/epoch - 6ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0048 - mse: 0.0048 - mae: 0.0534 - val_loss: 0.1081 - val_mse: 0.1081 - val_mae: 0.2871 - lr: 1.0000e-05 - 88ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0529 - val_loss: 0.1082 - val_mse: 0.1082 - val_mae: 0.2872 - lr: 1.0000e-05 - 91ms/epoch - 6ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0510 - val_loss: 0.1082 - val_mse: 0.1082 - val_mae: 0.2871 - lr: 1.0000e-05 - 92ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0534 - val_loss: 0.1081 - val_mse: 0.1081 - val_mae: 0.2871 - lr: 1.0000e-05 - 89ms/epoch - 6ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0512 - val_loss: 0.1082 - val_mse: 0.1082 - val_mae: 0.2871 - lr: 1.0000e-05 - 90ms/epoch - 6ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0555 - val_loss: 0.1082 - val_mse: 0.1082 - val_mae: 0.2872 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0521 - val_loss: 0.1082 - val_mse: 0.1082 - val_mae: 0.2872 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0048 - mse: 0.0048 - mae: 0.0545 - val_loss: 0.1081 - val_mse: 0.1081 - val_mae: 0.2871 - lr: 1.0000e-05 - 78ms/epoch - 5ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0048 - mse: 0.0048 - mae: 0.0521 - val_loss: 0.1080 - val_mse: 0.1080 - val_mae: 0.2870 - lr: 1.0000e-05 - 83ms/epoch - 5ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0045 - mse: 0.0045 - mae: 0.0522 - val_loss: 0.1079 - val_mse: 0.1079 - val_mae: 0.2868 - lr: 1.0000e-05 - 94ms/epoch - 6ms/step
Epoch 56/500
Epoch 00056: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0518 - val_loss: 0.1079 - val_mse: 0.1079 - val_mae: 0.2867 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 57/500
Epoch 00057: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0515 - val_loss: 0.1079 - val_mse: 0.1079 - val_mae: 0.2868 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 58/500
Epoch 00058: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0053 - mse: 0.0053 - mae: 0.0564 - val_loss: 0.1079 - val_mse: 0.1079 - val_mae: 0.2868 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 59/500
Epoch 00059: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0535 - val_loss: 0.1078 - val_mse: 0.1078 - val_mae: 0.2866 - lr: 1.0000e-05 - 84ms/epoch - 5ms/step
Epoch 60/500
Epoch 00060: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0049 - mse: 0.0049 - mae: 0.0549 - val_loss: 0.1077 - val_mse: 0.1077 - val_mae: 0.2865 - lr: 1.0000e-05 - 83ms/epoch - 5ms/step
Epoch 61/500
Epoch 00061: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0043 - mse: 0.0043 - mae: 0.0514 - val_loss: 0.1077 - val_mse: 0.1077 - val_mae: 0.2864 - lr: 1.0000e-05 - 92ms/epoch - 6ms/step
Epoch 62/500
Epoch 00062: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0511 - val_loss: 0.1077 - val_mse: 0.1077 - val_mae: 0.2864 - lr: 1.0000e-05 - 90ms/epoch - 6ms/step
Epoch 63/500
Epoch 00063: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0048 - mse: 0.0048 - mae: 0.0547 - val_loss: 0.1077 - val_mse: 0.1077 - val_mae: 0.2865 - lr: 1.0000e-05 - 79ms/epoch - 5ms/step
Epoch 64/500
Epoch 00064: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0522 - val_loss: 0.1077 - val_mse: 0.1077 - val_mae: 0.2865 - lr: 1.0000e-05 - 80ms/epoch - 5ms/step
Epoch 65/500
Epoch 00065: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0043 - mse: 0.0043 - mae: 0.0516 - val_loss: 0.1077 - val_mse: 0.1077 - val_mae: 0.2865 - lr: 1.0000e-05 - 88ms/epoch - 6ms/step
Epoch 66/500
Epoch 00066: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0524 - val_loss: 0.1077 - val_mse: 0.1077 - val_mae: 0.2865 - lr: 1.0000e-05 - 77ms/epoch - 5ms/step
Epoch 67/500
Epoch 00067: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0045 - mse: 0.0045 - mae: 0.0522 - val_loss: 0.1076 - val_mse: 0.1076 - val_mae: 0.2863 - lr: 1.0000e-05 - 88ms/epoch - 5ms/step
Epoch 68/500
Epoch 00068: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0521 - val_loss: 0.1076 - val_mse: 0.1076 - val_mae: 0.2863 - lr: 1.0000e-05 - 86ms/epoch - 5ms/step
Epoch 69/500
Epoch 00069: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0541 - val_loss: 0.1077 - val_mse: 0.1077 - val_mae: 0.2864 - lr: 1.0000e-05 - 87ms/epoch - 5ms/step
Epoch 70/500
Epoch 00070: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0526 - val_loss: 0.1076 - val_mse: 0.1076 - val_mae: 0.2863 - lr: 1.0000e-05 - 88ms/epoch - 6ms/step
Epoch 71/500
Epoch 00071: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0045 - mse: 0.0045 - mae: 0.0519 - val_loss: 0.1076 - val_mse: 0.1076 - val_mae: 0.2863 - lr: 1.0000e-05 - 92ms/epoch - 6ms/step
Epoch 72/500
Epoch 00072: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0537 - val_loss: 0.1074 - val_mse: 0.1074 - val_mae: 0.2860 - lr: 1.0000e-05 - 80ms/epoch - 5ms/step
Epoch 73/500
Epoch 00073: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0539 - val_loss: 0.1073 - val_mse: 0.1073 - val_mae: 0.2858 - lr: 1.0000e-05 - 88ms/epoch - 6ms/step
Epoch 74/500
Epoch 00074: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0518 - val_loss: 0.1073 - val_mse: 0.1073 - val_mae: 0.2858 - lr: 1.0000e-05 - 77ms/epoch - 5ms/step
Epoch 75/500
Epoch 00075: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0510 - val_loss: 0.1071 - val_mse: 0.1071 - val_mae: 0.2855 - lr: 1.0000e-05 - 90ms/epoch - 6ms/step
Epoch 76/500
Epoch 00076: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0043 - mse: 0.0043 - mae: 0.0517 - val_loss: 0.1070 - val_mse: 0.1070 - val_mae: 0.2853 - lr: 1.0000e-05 - 86ms/epoch - 5ms/step
Epoch 77/500
Epoch 00077: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0517 - val_loss: 0.1070 - val_mse: 0.1070 - val_mae: 0.2853 - lr: 1.0000e-05 - 86ms/epoch - 5ms/step
Epoch 78/500
Epoch 00078: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0533 - val_loss: 0.1069 - val_mse: 0.1069 - val_mae: 0.2851 - lr: 1.0000e-05 - 101ms/epoch - 6ms/step
Epoch 79/500
Epoch 00079: val_loss did not improve from 0.10331
16/16 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0532 - val_loss: 0.1069 - val_mse: 0.1069 - val_mae: 0.2852 - lr: 1.0000e-05 - 91ms/epoch - 6ms/step
Epoch 00079: early stopping
SMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 50.75% Accuracy
MSE: 39.71707698339657
RMSE: 6.302148600548591
MAPE: 5.205453851116191
EMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 18.546398408821787
RMSE: 4.306552961339474
MAPE: 3.4160340524918316
WMA
WMA([input_arrays], [timeperiod=30])
Weighted Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
49
Working on WMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.44 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4264.089, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3793.930, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.25 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3564.923, Time=0.07 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3427.258, Time=0.09 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.29 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.48 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3429.258, Time=0.18 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 2.877 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1709.629
Date: Sun, 12 Dec 2021 AIC 3427.258
Time: 14:19:34 BIC 3446.021
Sample: 0 HQIC 3434.464
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1981 0.003 -389.386 0.000 -1.204 -1.192
ar.L2 -0.8974 0.006 -139.699 0.000 -0.910 -0.885
ar.L3 -0.3983 0.006 -68.737 0.000 -0.410 -0.387
sigma2 4.0860 0.019 215.311 0.000 4.049 4.123
===================================================================================
Ljung-Box (L1) (Q): 14.57 Jarque-Bera (JB): 2460901.70
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.90
Prob(H) (two-sided): 0.00 Kurtosis: 273.75
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.40392, saving model to LSTM3.h5
17/17 - 3s - loss: 0.6042 - mse: 0.6042 - mae: 0.6609 - val_loss: 0.4039 - val_mse: 0.4039 - val_mae: 0.6126 - lr: 0.0010 - 3s/epoch - 158ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.40392 to 0.25737, saving model to LSTM3.h5
17/17 - 0s - loss: 0.1306 - mse: 0.1306 - mae: 0.3161 - val_loss: 0.2574 - val_mse: 0.2574 - val_mae: 0.4835 - lr: 0.0010 - 106ms/epoch - 6ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.25737
17/17 - 0s - loss: 0.0763 - mse: 0.0763 - mae: 0.2346 - val_loss: 0.2577 - val_mse: 0.2577 - val_mae: 0.4859 - lr: 0.0010 - 113ms/epoch - 7ms/step
Epoch 4/500
Epoch 00004: val_loss improved from 0.25737 to 0.19405, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0442 - mse: 0.0442 - mae: 0.1700 - val_loss: 0.1941 - val_mse: 0.1941 - val_mae: 0.4185 - lr: 0.0010 - 102ms/epoch - 6ms/step
Epoch 5/500
Epoch 00005: val_loss improved from 0.19405 to 0.13565, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0332 - mse: 0.0332 - mae: 0.1478 - val_loss: 0.1357 - val_mse: 0.1357 - val_mae: 0.3443 - lr: 0.0010 - 114ms/epoch - 7ms/step
Epoch 6/500
Epoch 00006: val_loss improved from 0.13565 to 0.10726, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0275 - mse: 0.0275 - mae: 0.1331 - val_loss: 0.1073 - val_mse: 0.1073 - val_mae: 0.3024 - lr: 0.0010 - 104ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: val_loss improved from 0.10726 to 0.08374, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0231 - mse: 0.0231 - mae: 0.1227 - val_loss: 0.0837 - val_mse: 0.0837 - val_mae: 0.2615 - lr: 0.0010 - 117ms/epoch - 7ms/step
Epoch 8/500
Epoch 00008: val_loss improved from 0.08374 to 0.06914, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0196 - mse: 0.0196 - mae: 0.1126 - val_loss: 0.0691 - val_mse: 0.0691 - val_mae: 0.2326 - lr: 0.0010 - 95ms/epoch - 6ms/step
Epoch 9/500
Epoch 00009: val_loss improved from 0.06914 to 0.06246, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0189 - mse: 0.0189 - mae: 0.1095 - val_loss: 0.0625 - val_mse: 0.0625 - val_mae: 0.2185 - lr: 0.0010 - 104ms/epoch - 6ms/step
Epoch 10/500
Epoch 00010: val_loss improved from 0.06246 to 0.05513, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0154 - mse: 0.0154 - mae: 0.0990 - val_loss: 0.0551 - val_mse: 0.0551 - val_mae: 0.2025 - lr: 0.0010 - 117ms/epoch - 7ms/step
Epoch 11/500
Epoch 00011: val_loss improved from 0.05513 to 0.04836, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0165 - mse: 0.0165 - mae: 0.1030 - val_loss: 0.0484 - val_mse: 0.0484 - val_mae: 0.1872 - lr: 0.0010 - 113ms/epoch - 7ms/step
Epoch 12/500
Epoch 00012: val_loss improved from 0.04836 to 0.04551, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0142 - mse: 0.0142 - mae: 0.0956 - val_loss: 0.0455 - val_mse: 0.0455 - val_mae: 0.1807 - lr: 0.0010 - 103ms/epoch - 6ms/step
Epoch 13/500
Epoch 00013: val_loss improved from 0.04551 to 0.04381, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0152 - mse: 0.0152 - mae: 0.0977 - val_loss: 0.0438 - val_mse: 0.0438 - val_mae: 0.1773 - lr: 0.0010 - 109ms/epoch - 6ms/step
Epoch 14/500
Epoch 00014: val_loss improved from 0.04381 to 0.04064, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0128 - mse: 0.0128 - mae: 0.0890 - val_loss: 0.0406 - val_mse: 0.0406 - val_mae: 0.1698 - lr: 0.0010 - 99ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.04064
17/17 - 0s - loss: 0.0114 - mse: 0.0114 - mae: 0.0860 - val_loss: 0.0407 - val_mse: 0.0407 - val_mae: 0.1710 - lr: 0.0010 - 94ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: val_loss improved from 0.04064 to 0.03795, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0118 - mse: 0.0118 - mae: 0.0862 - val_loss: 0.0380 - val_mse: 0.0380 - val_mae: 0.1645 - lr: 0.0010 - 115ms/epoch - 7ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.03795
17/17 - 0s - loss: 0.0110 - mse: 0.0110 - mae: 0.0818 - val_loss: 0.0381 - val_mse: 0.0381 - val_mae: 0.1654 - lr: 0.0010 - 103ms/epoch - 6ms/step
Epoch 18/500
Epoch 00018: val_loss improved from 0.03795 to 0.03378, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0093 - mse: 0.0093 - mae: 0.0759 - val_loss: 0.0338 - val_mse: 0.0338 - val_mae: 0.1538 - lr: 0.0010 - 103ms/epoch - 6ms/step
Epoch 19/500
Epoch 00019: val_loss improved from 0.03378 to 0.03132, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0091 - mse: 0.0091 - mae: 0.0743 - val_loss: 0.0313 - val_mse: 0.0313 - val_mae: 0.1473 - lr: 0.0010 - 99ms/epoch - 6ms/step
Epoch 20/500
Epoch 00020: val_loss improved from 0.03132 to 0.02844, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0089 - mse: 0.0089 - mae: 0.0758 - val_loss: 0.0284 - val_mse: 0.0284 - val_mae: 0.1391 - lr: 0.0010 - 118ms/epoch - 7ms/step
Epoch 21/500
Epoch 00021: val_loss improved from 0.02844 to 0.02538, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0091 - mse: 0.0091 - mae: 0.0744 - val_loss: 0.0254 - val_mse: 0.0254 - val_mae: 0.1298 - lr: 0.0010 - 137ms/epoch - 8ms/step
Epoch 22/500
Epoch 00022: val_loss improved from 0.02538 to 0.02217, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0083 - mse: 0.0083 - mae: 0.0729 - val_loss: 0.0222 - val_mse: 0.0222 - val_mae: 0.1196 - lr: 0.0010 - 100ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: val_loss improved from 0.02217 to 0.02136, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0080 - mse: 0.0080 - mae: 0.0700 - val_loss: 0.0214 - val_mse: 0.0214 - val_mae: 0.1171 - lr: 0.0010 - 109ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.02136
17/17 - 0s - loss: 0.0070 - mse: 0.0070 - mae: 0.0654 - val_loss: 0.0221 - val_mse: 0.0221 - val_mae: 0.1194 - lr: 0.0010 - 103ms/epoch - 6ms/step
Epoch 25/500
Epoch 00025: val_loss improved from 0.02136 to 0.01980, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0073 - mse: 0.0073 - mae: 0.0672 - val_loss: 0.0198 - val_mse: 0.0198 - val_mae: 0.1110 - lr: 0.0010 - 116ms/epoch - 7ms/step
Epoch 26/500
Epoch 00026: val_loss improved from 0.01980 to 0.01751, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0067 - mse: 0.0067 - mae: 0.0654 - val_loss: 0.0175 - val_mse: 0.0175 - val_mae: 0.1029 - lr: 0.0010 - 119ms/epoch - 7ms/step
Epoch 27/500
Epoch 00027: val_loss improved from 0.01751 to 0.01636, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0069 - mse: 0.0069 - mae: 0.0658 - val_loss: 0.0164 - val_mse: 0.0164 - val_mae: 0.0988 - lr: 0.0010 - 119ms/epoch - 7ms/step
Epoch 28/500
Epoch 00028: val_loss improved from 0.01636 to 0.01474, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0596 - val_loss: 0.0147 - val_mse: 0.0147 - val_mae: 0.0927 - lr: 0.0010 - 109ms/epoch - 6ms/step
Epoch 29/500
Epoch 00029: val_loss improved from 0.01474 to 0.01353, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0605 - val_loss: 0.0135 - val_mse: 0.0135 - val_mae: 0.0886 - lr: 0.0010 - 133ms/epoch - 8ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.01353
17/17 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0614 - val_loss: 0.0153 - val_mse: 0.0153 - val_mae: 0.0945 - lr: 0.0010 - 105ms/epoch - 6ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.01353
17/17 - 0s - loss: 0.0065 - mse: 0.0065 - mae: 0.0622 - val_loss: 0.0136 - val_mse: 0.0136 - val_mae: 0.0887 - lr: 0.0010 - 87ms/epoch - 5ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.01353
17/17 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0616 - val_loss: 0.0142 - val_mse: 0.0142 - val_mae: 0.0903 - lr: 0.0010 - 87ms/epoch - 5ms/step
Epoch 33/500
Epoch 00033: val_loss improved from 0.01353 to 0.01202, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0601 - val_loss: 0.0120 - val_mse: 0.0120 - val_mae: 0.0844 - lr: 0.0010 - 113ms/epoch - 7ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.01202
17/17 - 0s - loss: 0.0053 - mse: 0.0053 - mae: 0.0578 - val_loss: 0.0121 - val_mse: 0.0121 - val_mae: 0.0843 - lr: 0.0010 - 90ms/epoch - 5ms/step
Epoch 35/500
Epoch 00035: val_loss improved from 0.01202 to 0.01100, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0564 - val_loss: 0.0110 - val_mse: 0.0110 - val_mae: 0.0814 - lr: 0.0010 - 108ms/epoch - 6ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.01100
17/17 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0555 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0826 - lr: 0.0010 - 83ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.01100
17/17 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0528 - val_loss: 0.0120 - val_mse: 0.0120 - val_mae: 0.0831 - lr: 0.0010 - 92ms/epoch - 5ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.01100
17/17 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0553 - val_loss: 0.0143 - val_mse: 0.0143 - val_mae: 0.0894 - lr: 0.0010 - 91ms/epoch - 5ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.01100
17/17 - 0s - loss: 0.0050 - mse: 0.0050 - mae: 0.0542 - val_loss: 0.0119 - val_mse: 0.0119 - val_mae: 0.0829 - lr: 0.0010 - 101ms/epoch - 6ms/step
Epoch 40/500
Epoch 00040: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00040: val_loss did not improve from 0.01100
17/17 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0554 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0808 - lr: 0.0010 - 83ms/epoch - 5ms/step
Epoch 41/500
Epoch 00041: val_loss improved from 0.01100 to 0.01093, saving model to LSTM3.h5
17/17 - 0s - loss: 0.0050 - mse: 0.0050 - mae: 0.0542 - val_loss: 0.0109 - val_mse: 0.0109 - val_mae: 0.0805 - lr: 1.0000e-04 - 118ms/epoch - 7ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0530 - val_loss: 0.0110 - val_mse: 0.0110 - val_mae: 0.0806 - lr: 1.0000e-04 - 86ms/epoch - 5ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0531 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0809 - lr: 1.0000e-04 - 90ms/epoch - 5ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0514 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0811 - lr: 1.0000e-04 - 101ms/epoch - 6ms/step
Epoch 45/500
Epoch 00045: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00045: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0521 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0812 - lr: 1.0000e-04 - 93ms/epoch - 5ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0534 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0812 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0532 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0812 - lr: 1.0000e-05 - 92ms/epoch - 5ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0049 - mse: 0.0049 - mae: 0.0548 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0812 - lr: 1.0000e-05 - 96ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0528 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0813 - lr: 1.0000e-05 - 106ms/epoch - 6ms/step
Epoch 50/500
Epoch 00050: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00050: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0514 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0813 - lr: 1.0000e-05 - 88ms/epoch - 5ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0043 - mse: 0.0043 - mae: 0.0512 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0813 - lr: 1.0000e-05 - 91ms/epoch - 5ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0525 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0813 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0539 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0813 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0528 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0813 - lr: 1.0000e-05 - 96ms/epoch - 6ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0507 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0813 - lr: 1.0000e-05 - 84ms/epoch - 5ms/step
Epoch 56/500
Epoch 00056: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0045 - mse: 0.0045 - mae: 0.0518 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0814 - lr: 1.0000e-05 - 83ms/epoch - 5ms/step
Epoch 57/500
Epoch 00057: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0048 - mse: 0.0048 - mae: 0.0544 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0813 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 58/500
Epoch 00058: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0534 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0813 - lr: 1.0000e-05 - 96ms/epoch - 6ms/step
Epoch 59/500
Epoch 00059: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0528 - val_loss: 0.0114 - val_mse: 0.0114 - val_mae: 0.0814 - lr: 1.0000e-05 - 87ms/epoch - 5ms/step
Epoch 60/500
Epoch 00060: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0519 - val_loss: 0.0114 - val_mse: 0.0114 - val_mae: 0.0815 - lr: 1.0000e-05 - 88ms/epoch - 5ms/step
Epoch 61/500
Epoch 00061: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0515 - val_loss: 0.0114 - val_mse: 0.0114 - val_mae: 0.0815 - lr: 1.0000e-05 - 84ms/epoch - 5ms/step
Epoch 62/500
Epoch 00062: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0040 - mse: 0.0040 - mae: 0.0492 - val_loss: 0.0114 - val_mse: 0.0114 - val_mae: 0.0815 - lr: 1.0000e-05 - 81ms/epoch - 5ms/step
Epoch 63/500
Epoch 00063: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0530 - val_loss: 0.0114 - val_mse: 0.0114 - val_mae: 0.0815 - lr: 1.0000e-05 - 78ms/epoch - 5ms/step
Epoch 64/500
Epoch 00064: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0041 - mse: 0.0041 - mae: 0.0501 - val_loss: 0.0114 - val_mse: 0.0114 - val_mae: 0.0815 - lr: 1.0000e-05 - 81ms/epoch - 5ms/step
Epoch 65/500
Epoch 00065: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0528 - val_loss: 0.0114 - val_mse: 0.0114 - val_mae: 0.0815 - lr: 1.0000e-05 - 80ms/epoch - 5ms/step
Epoch 66/500
Epoch 00066: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0041 - mse: 0.0041 - mae: 0.0503 - val_loss: 0.0114 - val_mse: 0.0114 - val_mae: 0.0816 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 67/500
Epoch 00067: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0048 - mse: 0.0048 - mae: 0.0540 - val_loss: 0.0115 - val_mse: 0.0115 - val_mae: 0.0816 - lr: 1.0000e-05 - 84ms/epoch - 5ms/step
Epoch 68/500
Epoch 00068: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0498 - val_loss: 0.0115 - val_mse: 0.0115 - val_mae: 0.0816 - lr: 1.0000e-05 - 90ms/epoch - 5ms/step
Epoch 69/500
Epoch 00069: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0043 - mse: 0.0043 - mae: 0.0506 - val_loss: 0.0115 - val_mse: 0.0115 - val_mae: 0.0816 - lr: 1.0000e-05 - 92ms/epoch - 5ms/step
Epoch 70/500
Epoch 00070: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0505 - val_loss: 0.0115 - val_mse: 0.0115 - val_mae: 0.0816 - lr: 1.0000e-05 - 110ms/epoch - 6ms/step
Epoch 71/500
Epoch 00071: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0529 - val_loss: 0.0115 - val_mse: 0.0115 - val_mae: 0.0816 - lr: 1.0000e-05 - 91ms/epoch - 5ms/step
Epoch 72/500
Epoch 00072: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0499 - val_loss: 0.0115 - val_mse: 0.0115 - val_mae: 0.0816 - lr: 1.0000e-05 - 80ms/epoch - 5ms/step
Epoch 73/500
Epoch 00073: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0040 - mse: 0.0040 - mae: 0.0490 - val_loss: 0.0115 - val_mse: 0.0115 - val_mae: 0.0816 - lr: 1.0000e-05 - 82ms/epoch - 5ms/step
Epoch 74/500
Epoch 00074: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0506 - val_loss: 0.0115 - val_mse: 0.0115 - val_mae: 0.0817 - lr: 1.0000e-05 - 84ms/epoch - 5ms/step
Epoch 75/500
Epoch 00075: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0040 - mse: 0.0040 - mae: 0.0499 - val_loss: 0.0115 - val_mse: 0.0115 - val_mae: 0.0817 - lr: 1.0000e-05 - 87ms/epoch - 5ms/step
Epoch 76/500
Epoch 00076: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0528 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0818 - lr: 1.0000e-05 - 90ms/epoch - 5ms/step
Epoch 77/500
Epoch 00077: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0045 - mse: 0.0045 - mae: 0.0518 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0818 - lr: 1.0000e-05 - 82ms/epoch - 5ms/step
Epoch 78/500
Epoch 00078: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0043 - mse: 0.0043 - mae: 0.0512 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0819 - lr: 1.0000e-05 - 88ms/epoch - 5ms/step
Epoch 79/500
Epoch 00079: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0514 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0818 - lr: 1.0000e-05 - 88ms/epoch - 5ms/step
Epoch 80/500
Epoch 00080: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0513 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0819 - lr: 1.0000e-05 - 90ms/epoch - 5ms/step
Epoch 81/500
Epoch 00081: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0043 - mse: 0.0043 - mae: 0.0513 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0818 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 82/500
Epoch 00082: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0515 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0818 - lr: 1.0000e-05 - 89ms/epoch - 5ms/step
Epoch 83/500
Epoch 00083: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0045 - mse: 0.0045 - mae: 0.0521 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0818 - lr: 1.0000e-05 - 84ms/epoch - 5ms/step
Epoch 84/500
Epoch 00084: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0039 - mse: 0.0039 - mae: 0.0498 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0818 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 85/500
Epoch 00085: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0511 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0818 - lr: 1.0000e-05 - 86ms/epoch - 5ms/step
Epoch 86/500
Epoch 00086: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0048 - mse: 0.0048 - mae: 0.0538 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0818 - lr: 1.0000e-05 - 79ms/epoch - 5ms/step
Epoch 87/500
Epoch 00087: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0509 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0818 - lr: 1.0000e-05 - 84ms/epoch - 5ms/step
Epoch 88/500
Epoch 00088: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0048 - mse: 0.0048 - mae: 0.0550 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0818 - lr: 1.0000e-05 - 83ms/epoch - 5ms/step
Epoch 89/500
Epoch 00089: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0041 - mse: 0.0041 - mae: 0.0505 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0819 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 90/500
Epoch 00090: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0045 - mse: 0.0045 - mae: 0.0520 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0819 - lr: 1.0000e-05 - 96ms/epoch - 6ms/step
Epoch 91/500
Epoch 00091: val_loss did not improve from 0.01093
17/17 - 0s - loss: 0.0039 - mse: 0.0039 - mae: 0.0495 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0818 - lr: 1.0000e-05 - 100ms/epoch - 6ms/step
Epoch 00091: early stopping
SMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 50.75% Accuracy
MSE: 39.71707698339657
RMSE: 6.302148600548591
MAPE: 5.205453851116191
EMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 18.546398408821787
RMSE: 4.306552961339474
MAPE: 3.4160340524918316
WMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 43.66% Accuracy
MSE: 100.42411916020163
RMSE: 10.021183520932126
MAPE: 8.070563561893302
DEMA
DEMA([input_arrays], [timeperiod=30])
Double Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
89
Working on DEMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.44 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4436.126, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3965.317, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.40 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3736.589, Time=0.06 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3598.951, Time=0.09 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=0.98 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.98 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3600.951, Time=0.22 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.253 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1795.475
Date: Sun, 12 Dec 2021 AIC 3598.951
Time: 14:20:56 BIC 3617.714
Sample: 0 HQIC 3606.157
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1983 0.003 -389.581 0.000 -1.204 -1.192
ar.L2 -0.8973 0.006 -139.732 0.000 -0.910 -0.885
ar.L3 -0.3983 0.006 -68.649 0.000 -0.410 -0.387
sigma2 5.0573 0.023 215.292 0.000 5.011 5.103
===================================================================================
Ljung-Box (L1) (Q): 14.41 Jarque-Bera (JB): 2460553.80
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.89
Prob(H) (two-sided): 0.00 Kurtosis: 273.74
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.14773, saving model to LSTM3.h5
10/10 - 2s - loss: 0.3200 - mse: 0.3200 - mae: 0.4435 - val_loss: 0.1477 - val_mse: 0.1477 - val_mae: 0.3131 - lr: 0.0010 - 2s/epoch - 228ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.14773 to 0.11676, saving model to LSTM3.h5
10/10 - 0s - loss: 0.0717 - mse: 0.0717 - mae: 0.2237 - val_loss: 0.1168 - val_mse: 0.1168 - val_mae: 0.2746 - lr: 0.0010 - 81ms/epoch - 8ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0523 - mse: 0.0523 - mae: 0.1925 - val_loss: 0.1617 - val_mse: 0.1617 - val_mae: 0.3272 - lr: 0.0010 - 54ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0287 - mse: 0.0287 - mae: 0.1368 - val_loss: 0.2190 - val_mse: 0.2190 - val_mae: 0.3905 - lr: 0.0010 - 72ms/epoch - 7ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0301 - mse: 0.0301 - mae: 0.1390 - val_loss: 0.2329 - val_mse: 0.2329 - val_mae: 0.4074 - lr: 0.0010 - 51ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0244 - mse: 0.0244 - mae: 0.1254 - val_loss: 0.2363 - val_mse: 0.2363 - val_mae: 0.4131 - lr: 0.0010 - 63ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00007: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0192 - mse: 0.0192 - mae: 0.1103 - val_loss: 0.2484 - val_mse: 0.2484 - val_mae: 0.4273 - lr: 0.0010 - 58ms/epoch - 6ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0146 - mse: 0.0146 - mae: 0.0962 - val_loss: 0.2505 - val_mse: 0.2505 - val_mae: 0.4296 - lr: 1.0000e-04 - 58ms/epoch - 6ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0178 - mse: 0.0178 - mae: 0.1065 - val_loss: 0.2523 - val_mse: 0.2523 - val_mae: 0.4315 - lr: 1.0000e-04 - 56ms/epoch - 6ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0168 - mse: 0.0168 - mae: 0.1020 - val_loss: 0.2539 - val_mse: 0.2539 - val_mae: 0.4332 - lr: 1.0000e-04 - 60ms/epoch - 6ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0175 - mse: 0.0175 - mae: 0.1027 - val_loss: 0.2542 - val_mse: 0.2542 - val_mae: 0.4337 - lr: 1.0000e-04 - 65ms/epoch - 6ms/step
Epoch 12/500
Epoch 00012: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00012: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0161 - mse: 0.0161 - mae: 0.0996 - val_loss: 0.2549 - val_mse: 0.2549 - val_mae: 0.4346 - lr: 1.0000e-04 - 64ms/epoch - 6ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0156 - mse: 0.0156 - mae: 0.0996 - val_loss: 0.2547 - val_mse: 0.2547 - val_mae: 0.4345 - lr: 1.0000e-05 - 60ms/epoch - 6ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0157 - mse: 0.0157 - mae: 0.1004 - val_loss: 0.2547 - val_mse: 0.2547 - val_mae: 0.4344 - lr: 1.0000e-05 - 64ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0160 - mse: 0.0160 - mae: 0.0987 - val_loss: 0.2545 - val_mse: 0.2545 - val_mae: 0.4343 - lr: 1.0000e-05 - 59ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0159 - mse: 0.0159 - mae: 0.1002 - val_loss: 0.2542 - val_mse: 0.2542 - val_mae: 0.4340 - lr: 1.0000e-05 - 70ms/epoch - 7ms/step
Epoch 17/500
Epoch 00017: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00017: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0163 - mse: 0.0163 - mae: 0.1007 - val_loss: 0.2541 - val_mse: 0.2541 - val_mae: 0.4339 - lr: 1.0000e-05 - 82ms/epoch - 8ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0174 - mse: 0.0174 - mae: 0.1031 - val_loss: 0.2538 - val_mse: 0.2538 - val_mae: 0.4336 - lr: 1.0000e-05 - 68ms/epoch - 7ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0167 - mse: 0.0167 - mae: 0.1009 - val_loss: 0.2537 - val_mse: 0.2537 - val_mae: 0.4335 - lr: 1.0000e-05 - 57ms/epoch - 6ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0168 - mse: 0.0168 - mae: 0.1016 - val_loss: 0.2536 - val_mse: 0.2536 - val_mae: 0.4334 - lr: 1.0000e-05 - 71ms/epoch - 7ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0162 - mse: 0.0162 - mae: 0.1005 - val_loss: 0.2537 - val_mse: 0.2537 - val_mae: 0.4336 - lr: 1.0000e-05 - 66ms/epoch - 7ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0186 - mse: 0.0186 - mae: 0.1078 - val_loss: 0.2536 - val_mse: 0.2536 - val_mae: 0.4335 - lr: 1.0000e-05 - 59ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0157 - mse: 0.0157 - mae: 0.1002 - val_loss: 0.2537 - val_mse: 0.2537 - val_mae: 0.4336 - lr: 1.0000e-05 - 70ms/epoch - 7ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0171 - mse: 0.0171 - mae: 0.1028 - val_loss: 0.2538 - val_mse: 0.2538 - val_mae: 0.4337 - lr: 1.0000e-05 - 60ms/epoch - 6ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0165 - mse: 0.0165 - mae: 0.1022 - val_loss: 0.2541 - val_mse: 0.2541 - val_mae: 0.4341 - lr: 1.0000e-05 - 66ms/epoch - 7ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0179 - mse: 0.0179 - mae: 0.1043 - val_loss: 0.2544 - val_mse: 0.2544 - val_mae: 0.4344 - lr: 1.0000e-05 - 72ms/epoch - 7ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0163 - mse: 0.0163 - mae: 0.1028 - val_loss: 0.2547 - val_mse: 0.2547 - val_mae: 0.4348 - lr: 1.0000e-05 - 63ms/epoch - 6ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0161 - mse: 0.0161 - mae: 0.0988 - val_loss: 0.2549 - val_mse: 0.2549 - val_mae: 0.4350 - lr: 1.0000e-05 - 62ms/epoch - 6ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0164 - mse: 0.0164 - mae: 0.1008 - val_loss: 0.2550 - val_mse: 0.2550 - val_mae: 0.4351 - lr: 1.0000e-05 - 64ms/epoch - 6ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0171 - mse: 0.0171 - mae: 0.1019 - val_loss: 0.2550 - val_mse: 0.2550 - val_mae: 0.4352 - lr: 1.0000e-05 - 74ms/epoch - 7ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0159 - mse: 0.0159 - mae: 0.1002 - val_loss: 0.2551 - val_mse: 0.2551 - val_mae: 0.4352 - lr: 1.0000e-05 - 66ms/epoch - 7ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0151 - mse: 0.0151 - mae: 0.0987 - val_loss: 0.2551 - val_mse: 0.2551 - val_mae: 0.4353 - lr: 1.0000e-05 - 72ms/epoch - 7ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0151 - mse: 0.0151 - mae: 0.0983 - val_loss: 0.2551 - val_mse: 0.2551 - val_mae: 0.4353 - lr: 1.0000e-05 - 75ms/epoch - 8ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0152 - mse: 0.0152 - mae: 0.0968 - val_loss: 0.2549 - val_mse: 0.2549 - val_mae: 0.4351 - lr: 1.0000e-05 - 73ms/epoch - 7ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0164 - mse: 0.0164 - mae: 0.1015 - val_loss: 0.2548 - val_mse: 0.2548 - val_mae: 0.4351 - lr: 1.0000e-05 - 65ms/epoch - 6ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0164 - mse: 0.0164 - mae: 0.1019 - val_loss: 0.2549 - val_mse: 0.2549 - val_mae: 0.4351 - lr: 1.0000e-05 - 76ms/epoch - 8ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0164 - mse: 0.0164 - mae: 0.0988 - val_loss: 0.2550 - val_mse: 0.2550 - val_mae: 0.4353 - lr: 1.0000e-05 - 70ms/epoch - 7ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0154 - mse: 0.0154 - mae: 0.0979 - val_loss: 0.2552 - val_mse: 0.2552 - val_mae: 0.4356 - lr: 1.0000e-05 - 67ms/epoch - 7ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0153 - mse: 0.0153 - mae: 0.0978 - val_loss: 0.2554 - val_mse: 0.2554 - val_mae: 0.4357 - lr: 1.0000e-05 - 66ms/epoch - 7ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0170 - mse: 0.0170 - mae: 0.1026 - val_loss: 0.2553 - val_mse: 0.2553 - val_mae: 0.4357 - lr: 1.0000e-05 - 62ms/epoch - 6ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0163 - mse: 0.0163 - mae: 0.1024 - val_loss: 0.2555 - val_mse: 0.2555 - val_mae: 0.4359 - lr: 1.0000e-05 - 60ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0163 - mse: 0.0163 - mae: 0.1014 - val_loss: 0.2556 - val_mse: 0.2556 - val_mae: 0.4361 - lr: 1.0000e-05 - 54ms/epoch - 5ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0147 - mse: 0.0147 - mae: 0.0955 - val_loss: 0.2557 - val_mse: 0.2557 - val_mae: 0.4362 - lr: 1.0000e-05 - 60ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0162 - mse: 0.0162 - mae: 0.0995 - val_loss: 0.2557 - val_mse: 0.2557 - val_mae: 0.4362 - lr: 1.0000e-05 - 70ms/epoch - 7ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0167 - mse: 0.0167 - mae: 0.1036 - val_loss: 0.2557 - val_mse: 0.2557 - val_mae: 0.4362 - lr: 1.0000e-05 - 60ms/epoch - 6ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0165 - mse: 0.0165 - mae: 0.0992 - val_loss: 0.2556 - val_mse: 0.2556 - val_mae: 0.4361 - lr: 1.0000e-05 - 61ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0163 - mse: 0.0163 - mae: 0.0991 - val_loss: 0.2555 - val_mse: 0.2555 - val_mae: 0.4361 - lr: 1.0000e-05 - 68ms/epoch - 7ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0152 - mse: 0.0152 - mae: 0.0975 - val_loss: 0.2553 - val_mse: 0.2553 - val_mae: 0.4359 - lr: 1.0000e-05 - 69ms/epoch - 7ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0167 - mse: 0.0167 - mae: 0.1019 - val_loss: 0.2554 - val_mse: 0.2554 - val_mae: 0.4360 - lr: 1.0000e-05 - 60ms/epoch - 6ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0147 - mse: 0.0147 - mae: 0.0963 - val_loss: 0.2552 - val_mse: 0.2552 - val_mae: 0.4359 - lr: 1.0000e-05 - 53ms/epoch - 5ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0158 - mse: 0.0158 - mae: 0.0997 - val_loss: 0.2551 - val_mse: 0.2551 - val_mae: 0.4357 - lr: 1.0000e-05 - 57ms/epoch - 6ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.11676
10/10 - 0s - loss: 0.0154 - mse: 0.0154 - mae: 0.0962 - val_loss: 0.2551 - val_mse: 0.2551 - val_mae: 0.4358 - lr: 1.0000e-05 - 63ms/epoch - 6ms/step
Epoch 00052: early stopping
SMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 50.75% Accuracy
MSE: 39.71707698339657
RMSE: 6.302148600548591
MAPE: 5.205453851116191
EMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 18.546398408821787
RMSE: 4.306552961339474
MAPE: 3.4160340524918316
WMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 43.66% Accuracy
MSE: 100.42411916020163
RMSE: 10.021183520932126
MAPE: 8.070563561893302
DEMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 48.13% Accuracy
MSE: 42.724210554814
RMSE: 6.536375949623308
MAPE: 5.453197818871469
KAMA
KAMA([input_arrays], [timeperiod=30])
Kaufman Adaptive Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
18
Working on KAMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.34 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4190.464, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3724.371, Time=0.05 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.29 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3494.154, Time=0.08 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3357.435, Time=0.09 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.16 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.74 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3359.435, Time=0.19 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 2.983 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1674.717
Date: Sun, 12 Dec 2021 AIC 3357.435
Time: 14:22:08 BIC 3376.198
Sample: 0 HQIC 3364.641
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1955 0.003 -381.246 0.000 -1.202 -1.189
ar.L2 -0.8964 0.007 -135.835 0.000 -0.909 -0.883
ar.L3 -0.3971 0.006 -67.229 0.000 -0.409 -0.385
sigma2 3.7466 0.018 211.623 0.000 3.712 3.781
===================================================================================
Ljung-Box (L1) (Q): 14.20 Jarque-Bera (JB): 2338363.32
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.01 Skew: 3.76
Prob(H) (two-sided): 0.00 Kurtosis: 266.93
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.33764, saving model to LSTM3.h5
45/45 - 3s - loss: 0.2028 - mse: 0.2028 - mae: 0.3485 - val_loss: 0.3376 - val_mse: 0.3376 - val_mae: 0.5256 - lr: 0.0010 - 3s/epoch - 62ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.33764 to 0.18076, saving model to LSTM3.h5
45/45 - 0s - loss: 0.0340 - mse: 0.0340 - mae: 0.1503 - val_loss: 0.1808 - val_mse: 0.1808 - val_mae: 0.3622 - lr: 0.0010 - 238ms/epoch - 5ms/step
Epoch 3/500
Epoch 00003: val_loss improved from 0.18076 to 0.10560, saving model to LSTM3.h5
45/45 - 0s - loss: 0.0182 - mse: 0.0182 - mae: 0.1091 - val_loss: 0.1056 - val_mse: 0.1056 - val_mae: 0.2593 - lr: 0.0010 - 209ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss improved from 0.10560 to 0.07765, saving model to LSTM3.h5
45/45 - 0s - loss: 0.0144 - mse: 0.0144 - mae: 0.0962 - val_loss: 0.0777 - val_mse: 0.0777 - val_mae: 0.2164 - lr: 0.0010 - 233ms/epoch - 5ms/step
Epoch 5/500
Epoch 00005: val_loss improved from 0.07765 to 0.06452, saving model to LSTM3.h5
45/45 - 0s - loss: 0.0130 - mse: 0.0130 - mae: 0.0903 - val_loss: 0.0645 - val_mse: 0.0645 - val_mae: 0.1964 - lr: 0.0010 - 229ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0112 - mse: 0.0112 - mae: 0.0831 - val_loss: 0.0673 - val_mse: 0.0673 - val_mae: 0.2017 - lr: 0.0010 - 200ms/epoch - 4ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0110 - mse: 0.0110 - mae: 0.0820 - val_loss: 0.0738 - val_mse: 0.0738 - val_mae: 0.2127 - lr: 0.0010 - 211ms/epoch - 5ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0091 - mse: 0.0091 - mae: 0.0756 - val_loss: 0.0688 - val_mse: 0.0688 - val_mae: 0.2047 - lr: 0.0010 - 190ms/epoch - 4ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0086 - mse: 0.0086 - mae: 0.0732 - val_loss: 0.0700 - val_mse: 0.0700 - val_mae: 0.2064 - lr: 0.0010 - 193ms/epoch - 4ms/step
Epoch 10/500
Epoch 00010: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00010: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0078 - mse: 0.0078 - mae: 0.0697 - val_loss: 0.0715 - val_mse: 0.0715 - val_mae: 0.2094 - lr: 0.0010 - 235ms/epoch - 5ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0071 - mse: 0.0071 - mae: 0.0667 - val_loss: 0.0713 - val_mse: 0.0713 - val_mae: 0.2091 - lr: 1.0000e-04 - 210ms/epoch - 5ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0623 - val_loss: 0.0726 - val_mse: 0.0726 - val_mae: 0.2111 - lr: 1.0000e-04 - 181ms/epoch - 4ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0072 - mse: 0.0072 - mae: 0.0676 - val_loss: 0.0737 - val_mse: 0.0737 - val_mae: 0.2131 - lr: 1.0000e-04 - 187ms/epoch - 4ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0073 - mse: 0.0073 - mae: 0.0667 - val_loss: 0.0729 - val_mse: 0.0729 - val_mae: 0.2117 - lr: 1.0000e-04 - 187ms/epoch - 4ms/step
Epoch 15/500
Epoch 00015: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00015: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0632 - val_loss: 0.0741 - val_mse: 0.0741 - val_mae: 0.2137 - lr: 1.0000e-04 - 225ms/epoch - 5ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0628 - val_loss: 0.0742 - val_mse: 0.0742 - val_mae: 0.2139 - lr: 1.0000e-05 - 205ms/epoch - 5ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0622 - val_loss: 0.0743 - val_mse: 0.0743 - val_mae: 0.2141 - lr: 1.0000e-05 - 237ms/epoch - 5ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0645 - val_loss: 0.0745 - val_mse: 0.0745 - val_mae: 0.2144 - lr: 1.0000e-05 - 192ms/epoch - 4ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0065 - mse: 0.0065 - mae: 0.0644 - val_loss: 0.0745 - val_mse: 0.0745 - val_mae: 0.2145 - lr: 1.0000e-05 - 193ms/epoch - 4ms/step
Epoch 20/500
Epoch 00020: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00020: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0629 - val_loss: 0.0746 - val_mse: 0.0746 - val_mae: 0.2146 - lr: 1.0000e-05 - 192ms/epoch - 4ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0068 - mse: 0.0068 - mae: 0.0650 - val_loss: 0.0747 - val_mse: 0.0747 - val_mae: 0.2147 - lr: 1.0000e-05 - 216ms/epoch - 5ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0073 - mse: 0.0073 - mae: 0.0682 - val_loss: 0.0748 - val_mse: 0.0748 - val_mae: 0.2149 - lr: 1.0000e-05 - 176ms/epoch - 4ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0647 - val_loss: 0.0749 - val_mse: 0.0749 - val_mae: 0.2151 - lr: 1.0000e-05 - 213ms/epoch - 5ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0633 - val_loss: 0.0750 - val_mse: 0.0750 - val_mae: 0.2153 - lr: 1.0000e-05 - 215ms/epoch - 5ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0063 - mse: 0.0063 - mae: 0.0620 - val_loss: 0.0751 - val_mse: 0.0751 - val_mae: 0.2154 - lr: 1.0000e-05 - 195ms/epoch - 4ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0065 - mse: 0.0065 - mae: 0.0635 - val_loss: 0.0750 - val_mse: 0.0750 - val_mae: 0.2153 - lr: 1.0000e-05 - 218ms/epoch - 5ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0623 - val_loss: 0.0750 - val_mse: 0.0750 - val_mae: 0.2154 - lr: 1.0000e-05 - 212ms/epoch - 5ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0063 - mse: 0.0063 - mae: 0.0625 - val_loss: 0.0752 - val_mse: 0.0752 - val_mae: 0.2157 - lr: 1.0000e-05 - 212ms/epoch - 5ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0607 - val_loss: 0.0754 - val_mse: 0.0754 - val_mae: 0.2161 - lr: 1.0000e-05 - 213ms/epoch - 5ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0600 - val_loss: 0.0756 - val_mse: 0.0756 - val_mae: 0.2164 - lr: 1.0000e-05 - 224ms/epoch - 5ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0625 - val_loss: 0.0757 - val_mse: 0.0757 - val_mae: 0.2166 - lr: 1.0000e-05 - 216ms/epoch - 5ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0063 - mse: 0.0063 - mae: 0.0636 - val_loss: 0.0757 - val_mse: 0.0757 - val_mae: 0.2165 - lr: 1.0000e-05 - 181ms/epoch - 4ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0063 - mse: 0.0063 - mae: 0.0625 - val_loss: 0.0756 - val_mse: 0.0756 - val_mae: 0.2164 - lr: 1.0000e-05 - 184ms/epoch - 4ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0063 - mse: 0.0063 - mae: 0.0627 - val_loss: 0.0757 - val_mse: 0.0757 - val_mae: 0.2166 - lr: 1.0000e-05 - 180ms/epoch - 4ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0618 - val_loss: 0.0761 - val_mse: 0.0761 - val_mae: 0.2172 - lr: 1.0000e-05 - 190ms/epoch - 4ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0616 - val_loss: 0.0763 - val_mse: 0.0763 - val_mae: 0.2176 - lr: 1.0000e-05 - 198ms/epoch - 4ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0071 - mse: 0.0071 - mae: 0.0657 - val_loss: 0.0764 - val_mse: 0.0764 - val_mae: 0.2179 - lr: 1.0000e-05 - 182ms/epoch - 4ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0622 - val_loss: 0.0766 - val_mse: 0.0766 - val_mae: 0.2182 - lr: 1.0000e-05 - 213ms/epoch - 5ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0611 - val_loss: 0.0767 - val_mse: 0.0767 - val_mae: 0.2183 - lr: 1.0000e-05 - 222ms/epoch - 5ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0620 - val_loss: 0.0769 - val_mse: 0.0769 - val_mae: 0.2187 - lr: 1.0000e-05 - 192ms/epoch - 4ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0058 - mse: 0.0058 - mae: 0.0607 - val_loss: 0.0768 - val_mse: 0.0768 - val_mae: 0.2185 - lr: 1.0000e-05 - 189ms/epoch - 4ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0618 - val_loss: 0.0768 - val_mse: 0.0768 - val_mae: 0.2185 - lr: 1.0000e-05 - 181ms/epoch - 4ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0604 - val_loss: 0.0766 - val_mse: 0.0766 - val_mae: 0.2182 - lr: 1.0000e-05 - 194ms/epoch - 4ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0065 - mse: 0.0065 - mae: 0.0624 - val_loss: 0.0767 - val_mse: 0.0767 - val_mae: 0.2183 - lr: 1.0000e-05 - 184ms/epoch - 4ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0608 - val_loss: 0.0766 - val_mse: 0.0766 - val_mae: 0.2182 - lr: 1.0000e-05 - 193ms/epoch - 4ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0610 - val_loss: 0.0763 - val_mse: 0.0763 - val_mae: 0.2177 - lr: 1.0000e-05 - 176ms/epoch - 4ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0612 - val_loss: 0.0764 - val_mse: 0.0764 - val_mae: 0.2178 - lr: 1.0000e-05 - 219ms/epoch - 5ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0058 - mse: 0.0058 - mae: 0.0608 - val_loss: 0.0767 - val_mse: 0.0767 - val_mae: 0.2183 - lr: 1.0000e-05 - 219ms/epoch - 5ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0614 - val_loss: 0.0769 - val_mse: 0.0769 - val_mae: 0.2187 - lr: 1.0000e-05 - 187ms/epoch - 4ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0621 - val_loss: 0.0772 - val_mse: 0.0772 - val_mae: 0.2192 - lr: 1.0000e-05 - 231ms/epoch - 5ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0593 - val_loss: 0.0772 - val_mse: 0.0772 - val_mae: 0.2193 - lr: 1.0000e-05 - 199ms/epoch - 4ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0589 - val_loss: 0.0772 - val_mse: 0.0772 - val_mae: 0.2193 - lr: 1.0000e-05 - 186ms/epoch - 4ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0608 - val_loss: 0.0771 - val_mse: 0.0771 - val_mae: 0.2191 - lr: 1.0000e-05 - 221ms/epoch - 5ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0598 - val_loss: 0.0771 - val_mse: 0.0771 - val_mae: 0.2191 - lr: 1.0000e-05 - 214ms/epoch - 5ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.06452
45/45 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0577 - val_loss: 0.0771 - val_mse: 0.0771 - val_mae: 0.2190 - lr: 1.0000e-05 - 222ms/epoch - 5ms/step
Epoch 00055: early stopping
SMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 50.75% Accuracy
MSE: 39.71707698339657
RMSE: 6.302148600548591
MAPE: 5.205453851116191
EMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 18.546398408821787
RMSE: 4.306552961339474
MAPE: 3.4160340524918316
WMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 43.66% Accuracy
MSE: 100.42411916020163
RMSE: 10.021183520932126
MAPE: 8.070563561893302
DEMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 48.13% Accuracy
MSE: 42.724210554814
RMSE: 6.536375949623308
MAPE: 5.453197818871469
KAMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 23.65848136204503
RMSE: 4.863998495275778
MAPE: 3.972687129543795
MIDPOINT
MIDPOINT([input_arrays], [timeperiod=14])
MidPoint over period (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 14
Outputs:
real
14
Working on MIDPOINT predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.38 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4212.289, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3747.746, Time=0.05 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.24 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3523.401, Time=0.08 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3387.759, Time=0.09 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.18 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.85 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3389.758, Time=0.19 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.099 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1689.879
Date: Sun, 12 Dec 2021 AIC 3387.759
Time: 14:23:32 BIC 3406.522
Sample: 0 HQIC 3394.964
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1878 0.003 -345.315 0.000 -1.195 -1.181
ar.L2 -0.8876 0.007 -121.809 0.000 -0.902 -0.873
ar.L3 -0.3957 0.007 -60.127 0.000 -0.409 -0.383
sigma2 3.8904 0.020 193.404 0.000 3.851 3.930
===================================================================================
Ljung-Box (L1) (Q): 13.21 Jarque-Bera (JB): 1659080.01
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.08 Skew: 3.28
Prob(H) (two-sided): 0.00 Kurtosis: 225.31
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.03274, saving model to LSTM3.h5
58/58 - 3s - loss: 0.0795 - mse: 0.0795 - mae: 0.2101 - val_loss: 0.0327 - val_mse: 0.0327 - val_mae: 0.1384 - lr: 0.0010 - 3s/epoch - 44ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0109 - mse: 0.0109 - mae: 0.0826 - val_loss: 0.1007 - val_mse: 0.1007 - val_mae: 0.2732 - lr: 0.0010 - 242ms/epoch - 4ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0113 - mse: 0.0113 - mae: 0.0823 - val_loss: 0.0718 - val_mse: 0.0718 - val_mae: 0.2251 - lr: 0.0010 - 231ms/epoch - 4ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0075 - mse: 0.0075 - mae: 0.0687 - val_loss: 0.0948 - val_mse: 0.0948 - val_mae: 0.2657 - lr: 0.0010 - 260ms/epoch - 4ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0080 - mse: 0.0080 - mae: 0.0704 - val_loss: 0.0852 - val_mse: 0.0852 - val_mae: 0.2506 - lr: 0.0010 - 246ms/epoch - 4ms/step
Epoch 6/500
Epoch 00006: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00006: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0065 - mse: 0.0065 - mae: 0.0640 - val_loss: 0.1130 - val_mse: 0.1130 - val_mae: 0.2981 - lr: 0.0010 - 241ms/epoch - 4ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0138 - mse: 0.0138 - mae: 0.0942 - val_loss: 0.0711 - val_mse: 0.0711 - val_mae: 0.2266 - lr: 1.0000e-04 - 258ms/epoch - 4ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0083 - mse: 0.0083 - mae: 0.0725 - val_loss: 0.0735 - val_mse: 0.0735 - val_mae: 0.2307 - lr: 1.0000e-04 - 243ms/epoch - 4ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0089 - mse: 0.0089 - mae: 0.0752 - val_loss: 0.0698 - val_mse: 0.0698 - val_mae: 0.2234 - lr: 1.0000e-04 - 270ms/epoch - 5ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0082 - mse: 0.0082 - mae: 0.0724 - val_loss: 0.0706 - val_mse: 0.0706 - val_mae: 0.2248 - lr: 1.0000e-04 - 247ms/epoch - 4ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00011: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0082 - mse: 0.0082 - mae: 0.0726 - val_loss: 0.0663 - val_mse: 0.0663 - val_mae: 0.2166 - lr: 1.0000e-04 - 262ms/epoch - 5ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0068 - mse: 0.0068 - mae: 0.0652 - val_loss: 0.0669 - val_mse: 0.0669 - val_mae: 0.2177 - lr: 1.0000e-05 - 274ms/epoch - 5ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0070 - mse: 0.0070 - mae: 0.0653 - val_loss: 0.0672 - val_mse: 0.0672 - val_mae: 0.2183 - lr: 1.0000e-05 - 232ms/epoch - 4ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0063 - mse: 0.0063 - mae: 0.0624 - val_loss: 0.0675 - val_mse: 0.0675 - val_mae: 0.2188 - lr: 1.0000e-05 - 269ms/epoch - 5ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0654 - val_loss: 0.0677 - val_mse: 0.0677 - val_mae: 0.2191 - lr: 1.0000e-05 - 242ms/epoch - 4ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00016: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0068 - mse: 0.0068 - mae: 0.0650 - val_loss: 0.0681 - val_mse: 0.0681 - val_mae: 0.2199 - lr: 1.0000e-05 - 253ms/epoch - 4ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0067 - mse: 0.0067 - mae: 0.0642 - val_loss: 0.0681 - val_mse: 0.0681 - val_mae: 0.2200 - lr: 1.0000e-05 - 238ms/epoch - 4ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0067 - mse: 0.0067 - mae: 0.0649 - val_loss: 0.0682 - val_mse: 0.0682 - val_mae: 0.2202 - lr: 1.0000e-05 - 233ms/epoch - 4ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0629 - val_loss: 0.0686 - val_mse: 0.0686 - val_mae: 0.2208 - lr: 1.0000e-05 - 234ms/epoch - 4ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0628 - val_loss: 0.0683 - val_mse: 0.0683 - val_mae: 0.2203 - lr: 1.0000e-05 - 268ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0611 - val_loss: 0.0683 - val_mse: 0.0683 - val_mae: 0.2202 - lr: 1.0000e-05 - 240ms/epoch - 4ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0612 - val_loss: 0.0686 - val_mse: 0.0686 - val_mae: 0.2208 - lr: 1.0000e-05 - 234ms/epoch - 4ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0605 - val_loss: 0.0690 - val_mse: 0.0690 - val_mae: 0.2216 - lr: 1.0000e-05 - 256ms/epoch - 4ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0630 - val_loss: 0.0694 - val_mse: 0.0694 - val_mae: 0.2222 - lr: 1.0000e-05 - 276ms/epoch - 5ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0067 - mse: 0.0067 - mae: 0.0651 - val_loss: 0.0692 - val_mse: 0.0692 - val_mae: 0.2219 - lr: 1.0000e-05 - 250ms/epoch - 4ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0628 - val_loss: 0.0696 - val_mse: 0.0696 - val_mae: 0.2225 - lr: 1.0000e-05 - 263ms/epoch - 5ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0610 - val_loss: 0.0693 - val_mse: 0.0693 - val_mae: 0.2219 - lr: 1.0000e-05 - 268ms/epoch - 5ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0631 - val_loss: 0.0694 - val_mse: 0.0694 - val_mae: 0.2222 - lr: 1.0000e-05 - 268ms/epoch - 5ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0612 - val_loss: 0.0699 - val_mse: 0.0699 - val_mae: 0.2230 - lr: 1.0000e-05 - 243ms/epoch - 4ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0616 - val_loss: 0.0702 - val_mse: 0.0702 - val_mae: 0.2236 - lr: 1.0000e-05 - 241ms/epoch - 4ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0612 - val_loss: 0.0703 - val_mse: 0.0703 - val_mae: 0.2239 - lr: 1.0000e-05 - 247ms/epoch - 4ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0595 - val_loss: 0.0711 - val_mse: 0.0711 - val_mae: 0.2252 - lr: 1.0000e-05 - 298ms/epoch - 5ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0623 - val_loss: 0.0714 - val_mse: 0.0714 - val_mae: 0.2258 - lr: 1.0000e-05 - 264ms/epoch - 5ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0602 - val_loss: 0.0711 - val_mse: 0.0711 - val_mae: 0.2253 - lr: 1.0000e-05 - 249ms/epoch - 4ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0063 - mse: 0.0063 - mae: 0.0628 - val_loss: 0.0710 - val_mse: 0.0710 - val_mae: 0.2249 - lr: 1.0000e-05 - 265ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0628 - val_loss: 0.0709 - val_mse: 0.0709 - val_mae: 0.2248 - lr: 1.0000e-05 - 248ms/epoch - 4ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0594 - val_loss: 0.0713 - val_mse: 0.0713 - val_mae: 0.2255 - lr: 1.0000e-05 - 241ms/epoch - 4ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0577 - val_loss: 0.0715 - val_mse: 0.0715 - val_mae: 0.2259 - lr: 1.0000e-05 - 268ms/epoch - 5ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0607 - val_loss: 0.0709 - val_mse: 0.0709 - val_mae: 0.2248 - lr: 1.0000e-05 - 281ms/epoch - 5ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0596 - val_loss: 0.0705 - val_mse: 0.0705 - val_mae: 0.2238 - lr: 1.0000e-05 - 296ms/epoch - 5ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0590 - val_loss: 0.0707 - val_mse: 0.0707 - val_mae: 0.2243 - lr: 1.0000e-05 - 246ms/epoch - 4ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0611 - val_loss: 0.0708 - val_mse: 0.0708 - val_mae: 0.2244 - lr: 1.0000e-05 - 231ms/epoch - 4ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0592 - val_loss: 0.0713 - val_mse: 0.0713 - val_mae: 0.2254 - lr: 1.0000e-05 - 244ms/epoch - 4ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0607 - val_loss: 0.0705 - val_mse: 0.0705 - val_mae: 0.2237 - lr: 1.0000e-05 - 244ms/epoch - 4ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0563 - val_loss: 0.0710 - val_mse: 0.0710 - val_mae: 0.2248 - lr: 1.0000e-05 - 224ms/epoch - 4ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0606 - val_loss: 0.0712 - val_mse: 0.0712 - val_mae: 0.2252 - lr: 1.0000e-05 - 226ms/epoch - 4ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0058 - mse: 0.0058 - mae: 0.0600 - val_loss: 0.0715 - val_mse: 0.0715 - val_mae: 0.2256 - lr: 1.0000e-05 - 263ms/epoch - 5ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0614 - val_loss: 0.0713 - val_mse: 0.0713 - val_mae: 0.2252 - lr: 1.0000e-05 - 255ms/epoch - 4ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0595 - val_loss: 0.0718 - val_mse: 0.0718 - val_mae: 0.2263 - lr: 1.0000e-05 - 288ms/epoch - 5ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0560 - val_loss: 0.0718 - val_mse: 0.0718 - val_mae: 0.2261 - lr: 1.0000e-05 - 238ms/epoch - 4ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.03274
58/58 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0582 - val_loss: 0.0717 - val_mse: 0.0717 - val_mae: 0.2260 - lr: 1.0000e-05 - 241ms/epoch - 4ms/step
Epoch 00051: early stopping
SMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 50.75% Accuracy
MSE: 39.71707698339657
RMSE: 6.302148600548591
MAPE: 5.205453851116191
EMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 18.546398408821787
RMSE: 4.306552961339474
MAPE: 3.4160340524918316
WMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 43.66% Accuracy
MSE: 100.42411916020163
RMSE: 10.021183520932126
MAPE: 8.070563561893302
DEMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 48.13% Accuracy
MSE: 42.724210554814
RMSE: 6.536375949623308
MAPE: 5.453197818871469
KAMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 23.65848136204503
RMSE: 4.863998495275778
MAPE: 3.972687129543795
MIDPOINT
Prediction vs Close: 45.9% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 21.13528196098577
RMSE: 4.5973124715409295
MAPE: 3.8079919461917573
T3
T3([input_arrays], [timeperiod=5], [vfactor=0.7])
Triple Exponential Moving Average (T3) (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 5
vfactor: 0.7
Outputs:
real
19
Working on T3 predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.37 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4414.515, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3944.062, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.35 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3715.173, Time=0.07 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3577.471, Time=0.08 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.49 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.59 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3579.471, Time=0.20 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.249 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1784.736
Date: Sun, 12 Dec 2021 AIC 3577.471
Time: 14:24:58 BIC 3596.235
Sample: 0 HQIC 3584.677
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1982 0.003 -389.844 0.000 -1.204 -1.192
ar.L2 -0.8974 0.006 -139.861 0.000 -0.910 -0.885
ar.L3 -0.3983 0.006 -68.862 0.000 -0.410 -0.387
sigma2 4.9242 0.023 215.469 0.000 4.879 4.969
===================================================================================
Ljung-Box (L1) (Q): 14.55 Jarque-Bera (JB): 2468024.38
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.90
Prob(H) (two-sided): 0.00 Kurtosis: 274.15
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.03963, saving model to LSTM3.h5
43/43 - 2s - loss: 0.1598 - mse: 0.1598 - mae: 0.2724 - val_loss: 0.0396 - val_mse: 0.0396 - val_mae: 0.1735 - lr: 0.0010 - 2s/epoch - 55ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.03963 to 0.02142, saving model to LSTM3.h5
43/43 - 0s - loss: 0.0428 - mse: 0.0428 - mae: 0.1692 - val_loss: 0.0214 - val_mse: 0.0214 - val_mae: 0.1156 - lr: 0.0010 - 231ms/epoch - 5ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.02142
43/43 - 0s - loss: 0.0214 - mse: 0.0214 - mae: 0.1165 - val_loss: 0.0221 - val_mse: 0.0221 - val_mae: 0.1191 - lr: 0.0010 - 211ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.02142
43/43 - 0s - loss: 0.0116 - mse: 0.0116 - mae: 0.0849 - val_loss: 0.0245 - val_mse: 0.0245 - val_mae: 0.1290 - lr: 0.0010 - 226ms/epoch - 5ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.02142
43/43 - 0s - loss: 0.0104 - mse: 0.0104 - mae: 0.0807 - val_loss: 0.0249 - val_mse: 0.0249 - val_mae: 0.1309 - lr: 0.0010 - 186ms/epoch - 4ms/step
Epoch 6/500
Epoch 00006: val_loss improved from 0.02142 to 0.02114, saving model to LSTM3.h5
43/43 - 0s - loss: 0.0095 - mse: 0.0095 - mae: 0.0767 - val_loss: 0.0211 - val_mse: 0.0211 - val_mae: 0.1176 - lr: 0.0010 - 213ms/epoch - 5ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0084 - mse: 0.0084 - mae: 0.0722 - val_loss: 0.0247 - val_mse: 0.0247 - val_mae: 0.1308 - lr: 0.0010 - 193ms/epoch - 4ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0081 - mse: 0.0081 - mae: 0.0702 - val_loss: 0.0220 - val_mse: 0.0220 - val_mae: 0.1211 - lr: 0.0010 - 199ms/epoch - 5ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0067 - mse: 0.0067 - mae: 0.0647 - val_loss: 0.0248 - val_mse: 0.0248 - val_mae: 0.1307 - lr: 0.0010 - 210ms/epoch - 5ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0076 - mse: 0.0076 - mae: 0.0674 - val_loss: 0.0272 - val_mse: 0.0272 - val_mae: 0.1386 - lr: 0.0010 - 185ms/epoch - 4ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00011: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0072 - mse: 0.0072 - mae: 0.0665 - val_loss: 0.0265 - val_mse: 0.0265 - val_mae: 0.1360 - lr: 0.0010 - 182ms/epoch - 4ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0065 - mse: 0.0065 - mae: 0.0627 - val_loss: 0.0224 - val_mse: 0.0224 - val_mae: 0.1218 - lr: 1.0000e-04 - 188ms/epoch - 4ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0595 - val_loss: 0.0220 - val_mse: 0.0220 - val_mae: 0.1201 - lr: 1.0000e-04 - 192ms/epoch - 4ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0058 - mse: 0.0058 - mae: 0.0602 - val_loss: 0.0221 - val_mse: 0.0221 - val_mae: 0.1206 - lr: 1.0000e-04 - 189ms/epoch - 4ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0598 - val_loss: 0.0213 - val_mse: 0.0213 - val_mae: 0.1177 - lr: 1.0000e-04 - 188ms/epoch - 4ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00016: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0593 - val_loss: 0.0214 - val_mse: 0.0214 - val_mae: 0.1181 - lr: 1.0000e-04 - 241ms/epoch - 6ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0581 - val_loss: 0.0214 - val_mse: 0.0214 - val_mae: 0.1181 - lr: 1.0000e-05 - 230ms/epoch - 5ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0584 - val_loss: 0.0215 - val_mse: 0.0215 - val_mae: 0.1181 - lr: 1.0000e-05 - 219ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0570 - val_loss: 0.0214 - val_mse: 0.0214 - val_mae: 0.1179 - lr: 1.0000e-05 - 215ms/epoch - 5ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0053 - mse: 0.0053 - mae: 0.0573 - val_loss: 0.0214 - val_mse: 0.0214 - val_mae: 0.1180 - lr: 1.0000e-05 - 220ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00021: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0575 - val_loss: 0.0214 - val_mse: 0.0214 - val_mae: 0.1180 - lr: 1.0000e-05 - 226ms/epoch - 5ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0054 - mse: 0.0054 - mae: 0.0580 - val_loss: 0.0214 - val_mse: 0.0214 - val_mae: 0.1181 - lr: 1.0000e-05 - 211ms/epoch - 5ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0050 - mse: 0.0050 - mae: 0.0558 - val_loss: 0.0215 - val_mse: 0.0215 - val_mae: 0.1182 - lr: 1.0000e-05 - 233ms/epoch - 5ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0580 - val_loss: 0.0214 - val_mse: 0.0214 - val_mae: 0.1181 - lr: 1.0000e-05 - 183ms/epoch - 4ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0053 - mse: 0.0053 - mae: 0.0557 - val_loss: 0.0215 - val_mse: 0.0215 - val_mae: 0.1184 - lr: 1.0000e-05 - 204ms/epoch - 5ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0058 - mse: 0.0058 - mae: 0.0601 - val_loss: 0.0216 - val_mse: 0.0216 - val_mae: 0.1187 - lr: 1.0000e-05 - 213ms/epoch - 5ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0054 - mse: 0.0054 - mae: 0.0573 - val_loss: 0.0216 - val_mse: 0.0216 - val_mae: 0.1187 - lr: 1.0000e-05 - 191ms/epoch - 4ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0592 - val_loss: 0.0216 - val_mse: 0.0216 - val_mae: 0.1187 - lr: 1.0000e-05 - 190ms/epoch - 4ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0052 - mse: 0.0052 - mae: 0.0566 - val_loss: 0.0217 - val_mse: 0.0217 - val_mae: 0.1192 - lr: 1.0000e-05 - 183ms/epoch - 4ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0583 - val_loss: 0.0217 - val_mse: 0.0217 - val_mae: 0.1188 - lr: 1.0000e-05 - 185ms/epoch - 4ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0565 - val_loss: 0.0217 - val_mse: 0.0217 - val_mae: 0.1190 - lr: 1.0000e-05 - 207ms/epoch - 5ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0053 - mse: 0.0053 - mae: 0.0561 - val_loss: 0.0217 - val_mse: 0.0217 - val_mae: 0.1189 - lr: 1.0000e-05 - 192ms/epoch - 4ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0054 - mse: 0.0054 - mae: 0.0586 - val_loss: 0.0217 - val_mse: 0.0217 - val_mae: 0.1189 - lr: 1.0000e-05 - 206ms/epoch - 5ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0053 - mse: 0.0053 - mae: 0.0579 - val_loss: 0.0216 - val_mse: 0.0216 - val_mae: 0.1187 - lr: 1.0000e-05 - 215ms/epoch - 5ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0574 - val_loss: 0.0217 - val_mse: 0.0217 - val_mae: 0.1190 - lr: 1.0000e-05 - 181ms/epoch - 4ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0584 - val_loss: 0.0218 - val_mse: 0.0218 - val_mae: 0.1195 - lr: 1.0000e-05 - 223ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0050 - mse: 0.0050 - mae: 0.0548 - val_loss: 0.0219 - val_mse: 0.0219 - val_mae: 0.1195 - lr: 1.0000e-05 - 238ms/epoch - 6ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0585 - val_loss: 0.0219 - val_mse: 0.0219 - val_mae: 0.1198 - lr: 1.0000e-05 - 237ms/epoch - 6ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0050 - mse: 0.0050 - mae: 0.0543 - val_loss: 0.0220 - val_mse: 0.0220 - val_mae: 0.1199 - lr: 1.0000e-05 - 177ms/epoch - 4ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0052 - mse: 0.0052 - mae: 0.0568 - val_loss: 0.0220 - val_mse: 0.0220 - val_mae: 0.1200 - lr: 1.0000e-05 - 220ms/epoch - 5ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0544 - val_loss: 0.0220 - val_mse: 0.0220 - val_mae: 0.1199 - lr: 1.0000e-05 - 231ms/epoch - 5ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0578 - val_loss: 0.0219 - val_mse: 0.0219 - val_mae: 0.1196 - lr: 1.0000e-05 - 218ms/epoch - 5ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0050 - mse: 0.0050 - mae: 0.0553 - val_loss: 0.0218 - val_mse: 0.0218 - val_mae: 0.1195 - lr: 1.0000e-05 - 220ms/epoch - 5ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0053 - mse: 0.0053 - mae: 0.0574 - val_loss: 0.0218 - val_mse: 0.0218 - val_mae: 0.1193 - lr: 1.0000e-05 - 186ms/epoch - 4ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0594 - val_loss: 0.0218 - val_mse: 0.0218 - val_mae: 0.1194 - lr: 1.0000e-05 - 225ms/epoch - 5ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0574 - val_loss: 0.0218 - val_mse: 0.0218 - val_mae: 0.1195 - lr: 1.0000e-05 - 207ms/epoch - 5ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0052 - mse: 0.0052 - mae: 0.0553 - val_loss: 0.0217 - val_mse: 0.0217 - val_mae: 0.1191 - lr: 1.0000e-05 - 219ms/epoch - 5ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0556 - val_loss: 0.0217 - val_mse: 0.0217 - val_mae: 0.1190 - lr: 1.0000e-05 - 188ms/epoch - 4ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0052 - mse: 0.0052 - mae: 0.0567 - val_loss: 0.0217 - val_mse: 0.0217 - val_mae: 0.1188 - lr: 1.0000e-05 - 209ms/epoch - 5ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0050 - mse: 0.0050 - mae: 0.0552 - val_loss: 0.0216 - val_mse: 0.0216 - val_mae: 0.1185 - lr: 1.0000e-05 - 225ms/epoch - 5ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0563 - val_loss: 0.0215 - val_mse: 0.0215 - val_mae: 0.1182 - lr: 1.0000e-05 - 221ms/epoch - 5ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0053 - mse: 0.0053 - mae: 0.0567 - val_loss: 0.0214 - val_mse: 0.0214 - val_mae: 0.1179 - lr: 1.0000e-05 - 186ms/epoch - 4ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0054 - mse: 0.0054 - mae: 0.0562 - val_loss: 0.0214 - val_mse: 0.0214 - val_mae: 0.1176 - lr: 1.0000e-05 - 220ms/epoch - 5ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0588 - val_loss: 0.0215 - val_mse: 0.0215 - val_mae: 0.1182 - lr: 1.0000e-05 - 184ms/epoch - 4ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0049 - mse: 0.0049 - mae: 0.0547 - val_loss: 0.0215 - val_mse: 0.0215 - val_mae: 0.1183 - lr: 1.0000e-05 - 197ms/epoch - 5ms/step
Epoch 56/500
Epoch 00056: val_loss did not improve from 0.02114
43/43 - 0s - loss: 0.0052 - mse: 0.0052 - mae: 0.0560 - val_loss: 0.0216 - val_mse: 0.0216 - val_mae: 0.1186 - lr: 1.0000e-05 - 230ms/epoch - 5ms/step
Epoch 00056: early stopping
SMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 50.75% Accuracy
MSE: 39.71707698339657
RMSE: 6.302148600548591
MAPE: 5.205453851116191
EMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 18.546398408821787
RMSE: 4.306552961339474
MAPE: 3.4160340524918316
WMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 43.66% Accuracy
MSE: 100.42411916020163
RMSE: 10.021183520932126
MAPE: 8.070563561893302
DEMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 48.13% Accuracy
MSE: 42.724210554814
RMSE: 6.536375949623308
MAPE: 5.453197818871469
KAMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 23.65848136204503
RMSE: 4.863998495275778
MAPE: 3.972687129543795
MIDPOINT
Prediction vs Close: 45.9% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 21.13528196098577
RMSE: 4.5973124715409295
MAPE: 3.8079919461917573
T3
Prediction vs Close: 55.22% Accuracy
Prediction vs Prediction: 47.76% Accuracy
MSE: 101.89511494721017
RMSE: 10.09431101894578
MAPE: 8.111172771475383
TEMA
TEMA([input_arrays], [timeperiod=30])
Triple Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
9
Working on TEMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.49 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4352.703, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3889.412, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.26 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3689.930, Time=0.05 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3574.245, Time=0.08 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.21 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.83 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3576.245, Time=0.20 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.223 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1783.123
Date: Sun, 12 Dec 2021 AIC 3574.245
Time: 14:26:26 BIC 3593.008
Sample: 0 HQIC 3581.451
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1480 0.004 -302.430 0.000 -1.155 -1.141
ar.L2 -0.8300 0.008 -99.682 0.000 -0.846 -0.814
ar.L3 -0.3687 0.007 -50.527 0.000 -0.383 -0.354
sigma2 4.9055 0.028 175.970 0.000 4.851 4.960
===================================================================================
Ljung-Box (L1) (Q): 11.61 Jarque-Bera (JB): 1261976.58
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.16 Skew: 2.52
Prob(H) (two-sided): 0.00 Kurtosis: 196.90
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.16719, saving model to LSTM3.h5
90/90 - 3s - loss: 0.1229 - mse: 0.1229 - mae: 0.2524 - val_loss: 0.1672 - val_mse: 0.1672 - val_mae: 0.3380 - lr: 0.0010 - 3s/epoch - 34ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0169 - mse: 0.0169 - mae: 0.1022 - val_loss: 0.3016 - val_mse: 0.3016 - val_mae: 0.4812 - lr: 0.0010 - 352ms/epoch - 4ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0179 - mse: 0.0179 - mae: 0.1024 - val_loss: 0.2529 - val_mse: 0.2529 - val_mae: 0.4416 - lr: 0.0010 - 373ms/epoch - 4ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0126 - mse: 0.0126 - mae: 0.0882 - val_loss: 0.2506 - val_mse: 0.2506 - val_mae: 0.4436 - lr: 0.0010 - 373ms/epoch - 4ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0102 - mse: 0.0102 - mae: 0.0789 - val_loss: 0.2940 - val_mse: 0.2940 - val_mae: 0.4901 - lr: 0.0010 - 364ms/epoch - 4ms/step
Epoch 6/500
Epoch 00006: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00006: val_loss did not improve from 0.16719
90/90 - 1s - loss: 0.0109 - mse: 0.0109 - mae: 0.0817 - val_loss: 0.2786 - val_mse: 0.2786 - val_mae: 0.4787 - lr: 0.0010 - 546ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0127 - mse: 0.0127 - mae: 0.0905 - val_loss: 0.2438 - val_mse: 0.2438 - val_mae: 0.4423 - lr: 1.0000e-04 - 369ms/epoch - 4ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0078 - mse: 0.0078 - mae: 0.0676 - val_loss: 0.2522 - val_mse: 0.2522 - val_mae: 0.4509 - lr: 1.0000e-04 - 373ms/epoch - 4ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0071 - mse: 0.0071 - mae: 0.0668 - val_loss: 0.2652 - val_mse: 0.2652 - val_mae: 0.4639 - lr: 1.0000e-04 - 398ms/epoch - 4ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0067 - mse: 0.0067 - mae: 0.0658 - val_loss: 0.2694 - val_mse: 0.2694 - val_mae: 0.4680 - lr: 1.0000e-04 - 443ms/epoch - 5ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00011: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0071 - mse: 0.0071 - mae: 0.0662 - val_loss: 0.2746 - val_mse: 0.2746 - val_mae: 0.4732 - lr: 1.0000e-04 - 403ms/epoch - 4ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0070 - mse: 0.0070 - mae: 0.0642 - val_loss: 0.2746 - val_mse: 0.2746 - val_mae: 0.4732 - lr: 1.0000e-05 - 369ms/epoch - 4ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0069 - mse: 0.0069 - mae: 0.0642 - val_loss: 0.2742 - val_mse: 0.2742 - val_mae: 0.4728 - lr: 1.0000e-05 - 377ms/epoch - 4ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0065 - mse: 0.0065 - mae: 0.0629 - val_loss: 0.2763 - val_mse: 0.2763 - val_mae: 0.4749 - lr: 1.0000e-05 - 376ms/epoch - 4ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0631 - val_loss: 0.2776 - val_mse: 0.2776 - val_mae: 0.4762 - lr: 1.0000e-05 - 359ms/epoch - 4ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00016: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0067 - mse: 0.0067 - mae: 0.0639 - val_loss: 0.2778 - val_mse: 0.2778 - val_mae: 0.4764 - lr: 1.0000e-05 - 491ms/epoch - 5ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0063 - mse: 0.0063 - mae: 0.0621 - val_loss: 0.2802 - val_mse: 0.2802 - val_mae: 0.4787 - lr: 1.0000e-05 - 368ms/epoch - 4ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0622 - val_loss: 0.2810 - val_mse: 0.2810 - val_mae: 0.4795 - lr: 1.0000e-05 - 461ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0068 - mse: 0.0068 - mae: 0.0639 - val_loss: 0.2814 - val_mse: 0.2814 - val_mae: 0.4799 - lr: 1.0000e-05 - 366ms/epoch - 4ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0631 - val_loss: 0.2829 - val_mse: 0.2829 - val_mae: 0.4814 - lr: 1.0000e-05 - 449ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0058 - mse: 0.0058 - mae: 0.0602 - val_loss: 0.2838 - val_mse: 0.2838 - val_mae: 0.4823 - lr: 1.0000e-05 - 416ms/epoch - 5ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0626 - val_loss: 0.2836 - val_mse: 0.2836 - val_mae: 0.4821 - lr: 1.0000e-05 - 404ms/epoch - 4ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0613 - val_loss: 0.2830 - val_mse: 0.2830 - val_mae: 0.4815 - lr: 1.0000e-05 - 434ms/epoch - 5ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0607 - val_loss: 0.2849 - val_mse: 0.2849 - val_mae: 0.4834 - lr: 1.0000e-05 - 389ms/epoch - 4ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0597 - val_loss: 0.2851 - val_mse: 0.2851 - val_mae: 0.4836 - lr: 1.0000e-05 - 363ms/epoch - 4ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0610 - val_loss: 0.2849 - val_mse: 0.2849 - val_mae: 0.4835 - lr: 1.0000e-05 - 403ms/epoch - 4ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0614 - val_loss: 0.2842 - val_mse: 0.2842 - val_mae: 0.4828 - lr: 1.0000e-05 - 360ms/epoch - 4ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0067 - mse: 0.0067 - mae: 0.0640 - val_loss: 0.2825 - val_mse: 0.2825 - val_mae: 0.4812 - lr: 1.0000e-05 - 382ms/epoch - 4ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0627 - val_loss: 0.2836 - val_mse: 0.2836 - val_mae: 0.4822 - lr: 1.0000e-05 - 396ms/epoch - 4ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0608 - val_loss: 0.2850 - val_mse: 0.2850 - val_mae: 0.4836 - lr: 1.0000e-05 - 381ms/epoch - 4ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0585 - val_loss: 0.2865 - val_mse: 0.2865 - val_mae: 0.4851 - lr: 1.0000e-05 - 471ms/epoch - 5ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0605 - val_loss: 0.2867 - val_mse: 0.2867 - val_mae: 0.4853 - lr: 1.0000e-05 - 363ms/epoch - 4ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0058 - mse: 0.0058 - mae: 0.0581 - val_loss: 0.2877 - val_mse: 0.2877 - val_mae: 0.4863 - lr: 1.0000e-05 - 417ms/epoch - 5ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0572 - val_loss: 0.2879 - val_mse: 0.2879 - val_mae: 0.4865 - lr: 1.0000e-05 - 376ms/epoch - 4ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0577 - val_loss: 0.2909 - val_mse: 0.2909 - val_mae: 0.4895 - lr: 1.0000e-05 - 369ms/epoch - 4ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0579 - val_loss: 0.2901 - val_mse: 0.2901 - val_mae: 0.4887 - lr: 1.0000e-05 - 370ms/epoch - 4ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0596 - val_loss: 0.2905 - val_mse: 0.2905 - val_mae: 0.4891 - lr: 1.0000e-05 - 369ms/epoch - 4ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0058 - mse: 0.0058 - mae: 0.0611 - val_loss: 0.2903 - val_mse: 0.2903 - val_mae: 0.4889 - lr: 1.0000e-05 - 383ms/epoch - 4ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0583 - val_loss: 0.2881 - val_mse: 0.2881 - val_mae: 0.4868 - lr: 1.0000e-05 - 390ms/epoch - 4ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0586 - val_loss: 0.2886 - val_mse: 0.2886 - val_mae: 0.4873 - lr: 1.0000e-05 - 372ms/epoch - 4ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0054 - mse: 0.0054 - mae: 0.0569 - val_loss: 0.2896 - val_mse: 0.2896 - val_mae: 0.4883 - lr: 1.0000e-05 - 380ms/epoch - 4ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0053 - mse: 0.0053 - mae: 0.0576 - val_loss: 0.2900 - val_mse: 0.2900 - val_mae: 0.4887 - lr: 1.0000e-05 - 378ms/epoch - 4ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0053 - mse: 0.0053 - mae: 0.0565 - val_loss: 0.2929 - val_mse: 0.2929 - val_mae: 0.4914 - lr: 1.0000e-05 - 387ms/epoch - 4ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0053 - mse: 0.0053 - mae: 0.0568 - val_loss: 0.2931 - val_mse: 0.2931 - val_mae: 0.4917 - lr: 1.0000e-05 - 456ms/epoch - 5ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0052 - mse: 0.0052 - mae: 0.0571 - val_loss: 0.2935 - val_mse: 0.2935 - val_mae: 0.4920 - lr: 1.0000e-05 - 381ms/epoch - 4ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0584 - val_loss: 0.2950 - val_mse: 0.2950 - val_mae: 0.4936 - lr: 1.0000e-05 - 367ms/epoch - 4ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0054 - mse: 0.0054 - mae: 0.0581 - val_loss: 0.2982 - val_mse: 0.2982 - val_mae: 0.4966 - lr: 1.0000e-05 - 372ms/epoch - 4ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0054 - mse: 0.0054 - mae: 0.0570 - val_loss: 0.2977 - val_mse: 0.2977 - val_mae: 0.4961 - lr: 1.0000e-05 - 387ms/epoch - 4ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0563 - val_loss: 0.2970 - val_mse: 0.2970 - val_mae: 0.4956 - lr: 1.0000e-05 - 371ms/epoch - 4ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0571 - val_loss: 0.2977 - val_mse: 0.2977 - val_mae: 0.4963 - lr: 1.0000e-05 - 380ms/epoch - 4ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.16719
90/90 - 0s - loss: 0.0058 - mse: 0.0058 - mae: 0.0584 - val_loss: 0.2998 - val_mse: 0.2998 - val_mae: 0.4983 - lr: 1.0000e-05 - 402ms/epoch - 4ms/step
Epoch 00051: early stopping
SMA Prediction vs Close: 52.61% Accuracy Prediction vs Prediction: 50.75% Accuracy MSE: 39.71707698339657 RMSE: 6.302148600548591 MAPE: 5.205453851116191 EMA Prediction vs Close: 52.99% Accuracy Prediction vs Prediction: 46.64% Accuracy MSE: 18.546398408821787 RMSE: 4.306552961339474 MAPE: 3.4160340524918316 WMA Prediction vs Close: 54.48% Accuracy Prediction vs Prediction: 43.66% Accuracy MSE: 100.42411916020163 RMSE: 10.021183520932126 MAPE: 8.070563561893302 DEMA Prediction vs Close: 54.48% Accuracy Prediction vs Prediction: 48.13% Accuracy MSE: 42.724210554814 RMSE: 6.536375949623308 MAPE: 5.453197818871469 KAMA Prediction vs Close: 54.48% Accuracy Prediction vs Prediction: 48.51% Accuracy MSE: 23.65848136204503 RMSE: 4.863998495275778 MAPE: 3.972687129543795 MIDPOINT Prediction vs Close: 45.9% Accuracy Prediction vs Prediction: 45.9% Accuracy MSE: 21.13528196098577 RMSE: 4.5973124715409295 MAPE: 3.8079919461917573 T3 Prediction vs Close: 55.22% Accuracy Prediction vs Prediction: 47.76% Accuracy MSE: 101.89511494721017 RMSE: 10.09431101894578 MAPE: 8.111172771475383 TEMA Prediction vs Close: 49.25% Accuracy Prediction vs Prediction: 49.25% Accuracy MSE: 10.999022660597852 RMSE: 3.316477447623887 MAPE: 2.655228978622383 Runtime: mins: 11.426148450466659
Architecture Used¶
In [103]:
from google.colab import files
import cv2
uploaded = files.upload()
In [104]:
img = cv2.imread('Experiment3.png')
plt.figure(figsize=(20,10))
plt.axis("off")
plt.title('LSTM Architecture '+imgfile,fontsize=18)
plt.imshow(img)
Out[104]:
<matplotlib.image.AxesImage at 0x7f75dc03b110>
Model Plots¶
In [101]:
with open('simulation3_data.json') as json_file:
simulation3 = json.load(json_file)
fileimg = 'Experiment3'
In [102]:
for i in range(len(list(simulation3.keys()))):
SIM = list(simulation3.keys())[i]
plot_train(simulation3,SIM)
plot_test(simulation3,SIM)
----- Train RMSE for SMA ----- 9.349037175519484 ----- Train_MSE_LSTM for SMA ----- 87.40449610924534 ----- Train MAE LSTM for SMA ----- 8.179049661388795
----- Test RMSE for SMA----- 6.302148600548591 ----- Test_MSE_LSTM for SMA----- 39.71707698339657 ----- Test_MAE_LSTM for SMA----- 5.205453851116191
----- Train RMSE for EMA ----- 10.408056772244768 ----- Train_MSE_LSTM for EMA ----- 108.32764577427017 ----- Train MAE LSTM for EMA ----- 9.272168477176514
----- Test RMSE for EMA----- 4.306552961339474 ----- Test_MSE_LSTM for EMA----- 18.546398408821787 ----- Test_MAE_LSTM for EMA----- 3.4160340524918316
----- Train RMSE for WMA ----- 10.758673148854621 ----- Train_MSE_LSTM for WMA ----- 115.7490479238854 ----- Train MAE LSTM for WMA ----- 9.635844746281826
----- Test RMSE for WMA----- 10.021183520932126 ----- Test_MSE_LSTM for WMA----- 100.42411916020163 ----- Test_MAE_LSTM for WMA----- 8.070563561893302
----- Train RMSE for DEMA ----- 12.471033474287747 ----- Train_MSE_LSTM for DEMA ----- 155.52667591680552 ----- Train MAE LSTM for DEMA ----- 11.27911882792913
----- Test RMSE for DEMA----- 6.536375949623308 ----- Test_MSE_LSTM for DEMA----- 42.724210554814 ----- Test_MAE_LSTM for DEMA----- 5.453197818871469
----- Train RMSE for KAMA ----- 11.226948346371925 ----- Train_MSE_LSTM for KAMA ----- 126.0443691721033 ----- Train MAE LSTM for KAMA ----- 10.243495551915379
----- Test RMSE for KAMA----- 4.863998495275778 ----- Test_MSE_LSTM for KAMA----- 23.65848136204503 ----- Test_MAE_LSTM for KAMA----- 3.972687129543795
----- Train RMSE for MIDPOINT ----- 9.85503947109961 ----- Train_MSE_LSTM for MIDPOINT ----- 97.1218029769313 ----- Train MAE LSTM for MIDPOINT ----- 8.735687748390125
----- Test RMSE for MIDPOINT----- 4.5973124715409295 ----- Test_MSE_LSTM for MIDPOINT----- 21.13528196098577 ----- Test_MAE_LSTM for MIDPOINT----- 3.8079919461917573
----- Train RMSE for T3 ----- 12.321219257781456 ----- Train_MSE_LSTM for T3 ----- 151.8124439983246 ----- Train MAE LSTM for T3 ----- 11.208239243433828
----- Test RMSE for T3----- 10.09431101894578 ----- Test_MSE_LSTM for T3----- 101.89511494721017 ----- Test_MAE_LSTM for T3----- 8.111172771475383
----- Train RMSE for TEMA ----- 7.47666641398143 ----- Train_MSE_LSTM for TEMA ----- 55.900540665957934 ----- Train MAE LSTM for TEMA ----- 5.218204706016292
----- Test RMSE for TEMA----- 3.316477447623887 ----- Test_MSE_LSTM for TEMA----- 10.999022660597852 ----- Test_MAE_LSTM for TEMA----- 2.655228978622383
Univariate Arima Multistep MutiVariate LSTM Hybrid Model Experiment 4¶
From the above experiments it is evident that with Higher moving averages the loss plots show unreoresented data and underfitting, hence keeping only the MA's that have smaller periods like T3 OR TRIMA. Going forward EMA, WMA & DEMA will be ignored.
In [106]:
def get_lstm(data,original_data, train_len, test_len,img_file,ma ,lstm_len=3):
# prepare train and test data
X_value = pd.DataFrame(data.iloc[:, :])
y_value = pd.DataFrame(data.iloc[:, 3])
X_scaler = MinMaxScaler(feature_range=(-1, 1))
y_scaler = MinMaxScaler(feature_range=(-1, 1))
X_scaler.fit(X_value)
y_scaler.fit(y_value)
# Get data and check shape
X, y, yc = get_X_y(X_scale_dataset, y_scale_dataset)#X will be of shape 224 X 3 X 21 (each 3 X 21 array will be 3 days' worth of data). yc will have the corresponding closing price value
# pdb.set_trace()
X_train, X_test, = split_train_test(X)
y_train, y_test, = split_train_test(y)
# yc_train, yc_test, = split_train_test(original_data)
index_train, index_test, = predict_index(dataset_final, X_train, n_steps_in, n_steps_out)
det = 20
input_dim = X_train.shape[1]#3
feature_size = X_train.shape[2]#24
output_dim = y_train.shape[1]#1
# # Option 1
# # Set up & fit LSTM RNN
# model = Sequential()
# model.add(LSTM(256, activation='relu', kernel_initializer='he_normal', input_shape=(input_dim, feature_size)))
# model.add(Dense(units=64,activation='relu'))
# model.add(Dropout(0.5))
# model.add(Dense(units=output_dim))
# model.compile(optimizer=Adam(learning_rate = 0.001), loss='mse')
# ## Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# # option 2
# model = Sequential()
# model.add(Bidirectional(LSTM(units= 128), input_shape=(input_dim, feature_size)))
# model.add(Dense(64))
# model.add(Dense(units=output_dim))
# model.compile(optimizer=Adam(lr = 0.001), loss='mean_squared_error', metrics=['accuracy'])
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# # Option 3
# # define custom activation
# #
# class Double_Tanh(Activation):
# def __init__(self, activation, **kwargs):
# super(Double_Tanh, self).__init__(activation, **kwargs)
# self.__name__ = 'double_tanh'
# def double_tanh(x):
# return (K.tanh(x) * 2)
# get_custom_objects().update({'double_tanh':Double_Tanh(double_tanh)})
# # Model Generation
# model = Sequential()
# #check https://machinelearningmastery.com/use-weight-regularization-lstm-networks-time-series-forecasting/
# model.add(LSTM(25, input_shape=(input_dim, feature_size), dropout=0.2, kernel_regularizer=l1_l2(0.00,0.00), bias_regularizer=l1_l2(0.00,0.00)))
# model.add(Dense(1))
# model.add(Activation(double_tanh))
# model.compile(loss='mean_squared_error', optimizer='adam', metrics=['mse', 'mae'])
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
pyplot.show()
# Option 4
# Set up & fit LSTM RNN
model = Sequential()
model.add(LSTM(units=lstm_len, return_sequences=True, input_shape=(input_dim, feature_size)))
model.add(LSTM(units=int(lstm_len/2)))
model.add(Dense(1, activation='sigmoid'))
model.compile(loss='mean_squared_error', optimizer='adam')
# Common code
callbacks = [
EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
ModelCheckpoint('LSTM4.h5', verbose=1, save_best_only=True, save_weights_only=True)]
fname1 = img_file+'.png'
tensorflow.keras.utils.plot_model(
model, to_file=fname1, show_shapes=True, show_dtype=False,
show_layer_names=True, expand_nested=False, dpi=96,
layer_range=None, show_layer_activations=False
)
history = model.fit(X_train, y_train, epochs=500, batch_size=int( optimized_period[ma]), verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# plot loss
fname2 = img_file+'-'+ma
plt.title(img_file+'-'+ma+' Loss')
plt.xlabel("Epochs")
plt.ylabel("Loss")
pyplot.plot(history.history['loss'], label='train')
pyplot.plot(history.history['val_loss'], label='validation')
pyplot.legend()
pyplot.savefig(fname2+'.png',dpi='figure')
pyplot.show()
# Generate predictions
predictiontr = model.predict(X_train, verbose=0)
predictiontr = (y_scaler.inverse_transform(predictiontr)-det).tolist()
outputtr = []
for i in range(len(predictiontr)):
outputtr.extend(predictiontr[i])
predictiontr = outputtr
# Generate error data
## replace with yc , xtest generated by new multistep method
mse_tr = mean_squared_error(y_train, predictiontr)
rmse_tr = mse_tr ** 0.5
# mape = mean_absolute_percentage_error(X_test, pd.Series(predictiontr))
mae_tr = mean_absolute_error(y_train, pd.Series(predictiontr))
# Original_tr = pd.Series(yc_train)
Original_tr = y_scaler.inverse_transform(y_train).flatten().tolist()
predictionte = model.predict(X_test, verbose=0)
predictionte =( y_scaler.inverse_transform(predictionte)-det).tolist()
outputte = []
for i in range(len(predictionte)):
outputte.extend(predictionte[i])
predictionte = outputte
# Generate error data
mse_te = mean_squared_error(y_test, predictionte)
rmse_te = mse_te ** 0.5
# mape = mean_absolute_percentage_error(X_test, pd.Series(predictionte))
mae_te = mean_absolute_error(y_test, pd.Series(predictionte))
# Original_te = pd.Series(yc_test)
Original_te = y_scaler.inverse_transform(y_test).flatten().tolist()
return Original_tr, predictiontr, mse_tr, rmse_tr,mae_tr,Original_te,predictionte, mse_te,rmse_te,mae_te
In [107]:
if __name__ == '__main__':
start_time = timeit.default_timer()
simulation4 = {}
imgfile = 'Experiment4'
for ma in optimized_period:
print(ma)
print(functions[ma])
print ( int( optimized_period[ma]))
# if ma == 'SMA':
low_vol = df.apply(lambda c: functions[ma](c, timeperiod = int( optimized_period[ma])))
low_vol = low_vol.fillna(0)
low_vol_data = df['close']
high_vol = pd.DataFrame()
df2 = df.copy()
for i in df2.columns:
if i in low_vol.columns:
high_vol[i] = df2[i].subtract(low_vol[i], fill_value=0)
high_vol_data = df['close']
## *****************************************************
# Generate ARIMA and LSTM predictions
print('\nWorking on ' + ma + ' predictions')
try:
print('parameters used : ', train_len, test_len)
low_vol_Original, low_vol_prediction, low_vol_mse, low_vol_rmse,low_vol_mae = get_arima(low_vol,low_vol_data, train_len, test_len)
except:
print('ARIMA error, skipping to next MA type')
continue
Original_tr, predictiontr, mse_tr, rmse_tr,mae_tr,high_vol_Original, high_vol_prediction, high_vol_mse, high_vol_rmse,high_vol_mae, = get_lstm(high_vol,high_vol_data, train_len, test_len,imgfile,ma)
final_prediction_tr = df['close'].head(train_len).values + pd.Series(predictiontr) # ignoring first 3 steps
mse_ftr = mean_squared_error(df['close'].head(train_len).values,final_prediction_tr.values)
rmse_ftr = mse_ftr ** 0.5
mape_ftr = mean_absolute_percentage_error(df['close'].head(train_len).reset_index(drop=True), final_prediction_tr)
mae_ftr = mean_absolute_error(df['close'].head(train_len).reset_index(drop=True), final_prediction_tr)
final_prediction = pd.Series(low_vol_prediction[3:]) + pd.Series(high_vol_prediction)
mse = mean_squared_error(df['close'].tail(test_len).values,final_prediction.values)
rmse = mse ** 0.5
mape = mean_absolute_percentage_error(df['close'].tail(test_len).reset_index(drop=True), final_prediction)
mae = mean_absolute_error(df['close'].tail(test_len).reset_index(drop=True), final_prediction)
# Generate prediction accuracy
actual = df['close'].tail(test_len).values
result_1 = []
result_2 = []
for i in range(1, len(final_prediction)):
# Compare prediction to previous close price
if final_prediction[i] > actual[i-1] and actual[i] > actual[i-1]:
result_1.append(1)
elif final_prediction[i] < actual[i-1] and actual[i] < actual[i-1]:
result_1.append(1)
else:
result_1.append(0)
# Compare prediction to previous prediction
if final_prediction[i] > final_prediction[i-1] and actual[i] > actual[i-1]:
result_2.append(1)
elif final_prediction[i] < final_prediction[i-1] and actual[i] < actual[i-1]:
result_2.append(1)
else:
result_2.append(0)
accuracy_1 = np.mean(result_1)
accuracy_2 = np.mean(result_2)
simulation4[ma] = {'low_vol': {'original':list(low_vol_Original), 'prediction': list(low_vol_prediction) , 'mse': low_vol_mse,
'rmse': low_vol_rmse, 'mae' : low_vol_mae},
'high_vol': {'original':list(high_vol_Original),'prediction': list(high_vol_prediction), 'mse': high_vol_mse,
'rmse': high_vol_rmse, 'mae' : high_vol_mae},
'final_tr': {'original':df['close'].head(train_len).tolist(),'prediction': final_prediction_tr.values.tolist(), 'mse': mse_ftr,
'rmse': rmse_ftr, 'mae' : mae_ftr},
'final': {'original': df['close'].tail(test_len).tolist(), 'prediction': final_prediction.values.tolist(), 'mse': mse,
'rmse': rmse, 'mae': mae },
'accuracy': {'prediction vs close': accuracy_1, 'prediction vs prediction': accuracy_2}}
# save simulation data here as checkpoint
with open('simulation4_data.json', 'w') as fp:
json.dump(simulation4, fp)
for ma in simulation4.keys():
print('\n' + ma)
print('Prediction vs Close:\t\t' + str(round(100*simulation4[ma]['accuracy']['prediction vs close'], 2))
+ '% Accuracy')
print('Prediction vs Prediction:\t' + str(round(100*simulation4[ma]['accuracy']['prediction vs prediction'], 2))
+ '% Accuracy')
print('MSE:\t', simulation4[ma]['final']['mse'],
'\nRMSE:\t', simulation4[ma]['final']['rmse'],
'\nMAPE:\t', simulation4[ma]['final']['mae'])#,
# '\nMAPE:\t', simulation[ma]['final']['mape'])
# else:
# break
elapsed = timeit.default_timer() - start_time
print('Runtime: mins:',elapsed/60)
SMA
SMA([input_arrays], [timeperiod=30])
Simple Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
17
Working on SMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.45 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4157.020, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3687.148, Time=0.05 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.21 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3458.651, Time=0.09 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3322.133, Time=0.11 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=0.75 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.80 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3324.133, Time=0.20 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 2.697 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1657.067
Date: Sun, 12 Dec 2021 AIC 3322.133
Time: 14:39:08 BIC 3340.897
Sample: 0 HQIC 3329.339
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1966 0.003 -387.226 0.000 -1.203 -1.191
ar.L2 -0.8952 0.006 -138.692 0.000 -0.908 -0.883
ar.L3 -0.3968 0.006 -68.284 0.000 -0.408 -0.385
sigma2 3.5858 0.017 214.535 0.000 3.553 3.619
===================================================================================
Ljung-Box (L1) (Q): 14.47 Jarque-Bera (JB): 2428881.42
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.90
Prob(H) (two-sided): 0.00 Kurtosis: 271.99
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.05005, saving model to LSTM4.h5
48/48 - 4s - loss: 1.3045 - val_loss: 0.0501 - lr: 0.0010 - 4s/epoch - 80ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.05005
48/48 - 0s - loss: 1.2224 - val_loss: 0.0536 - lr: 0.0010 - 281ms/epoch - 6ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.05005
48/48 - 0s - loss: 1.1389 - val_loss: 0.0565 - lr: 0.0010 - 243ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.05005
48/48 - 0s - loss: 1.0538 - val_loss: 0.0591 - lr: 0.0010 - 243ms/epoch - 5ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.9757 - val_loss: 0.0613 - lr: 0.0010 - 261ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00006: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.9088 - val_loss: 0.0639 - lr: 0.0010 - 283ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8728 - val_loss: 0.0642 - lr: 1.0000e-04 - 269ms/epoch - 6ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8678 - val_loss: 0.0645 - lr: 1.0000e-04 - 292ms/epoch - 6ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8628 - val_loss: 0.0649 - lr: 1.0000e-04 - 278ms/epoch - 6ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8579 - val_loss: 0.0653 - lr: 1.0000e-04 - 275ms/epoch - 6ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00011: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8531 - val_loss: 0.0656 - lr: 1.0000e-04 - 246ms/epoch - 5ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8501 - val_loss: 0.0657 - lr: 1.0000e-05 - 247ms/epoch - 5ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8496 - val_loss: 0.0657 - lr: 1.0000e-05 - 271ms/epoch - 6ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8492 - val_loss: 0.0658 - lr: 1.0000e-05 - 290ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8487 - val_loss: 0.0658 - lr: 1.0000e-05 - 281ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00016: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8482 - val_loss: 0.0659 - lr: 1.0000e-05 - 254ms/epoch - 5ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8477 - val_loss: 0.0659 - lr: 1.0000e-05 - 292ms/epoch - 6ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8472 - val_loss: 0.0659 - lr: 1.0000e-05 - 237ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8467 - val_loss: 0.0660 - lr: 1.0000e-05 - 240ms/epoch - 5ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8462 - val_loss: 0.0660 - lr: 1.0000e-05 - 258ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8457 - val_loss: 0.0661 - lr: 1.0000e-05 - 277ms/epoch - 6ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8452 - val_loss: 0.0661 - lr: 1.0000e-05 - 246ms/epoch - 5ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8447 - val_loss: 0.0662 - lr: 1.0000e-05 - 268ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8442 - val_loss: 0.0662 - lr: 1.0000e-05 - 251ms/epoch - 5ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8437 - val_loss: 0.0663 - lr: 1.0000e-05 - 273ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8432 - val_loss: 0.0663 - lr: 1.0000e-05 - 240ms/epoch - 5ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8427 - val_loss: 0.0664 - lr: 1.0000e-05 - 249ms/epoch - 5ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8422 - val_loss: 0.0664 - lr: 1.0000e-05 - 234ms/epoch - 5ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8417 - val_loss: 0.0665 - lr: 1.0000e-05 - 243ms/epoch - 5ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8412 - val_loss: 0.0665 - lr: 1.0000e-05 - 248ms/epoch - 5ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8407 - val_loss: 0.0666 - lr: 1.0000e-05 - 273ms/epoch - 6ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8402 - val_loss: 0.0666 - lr: 1.0000e-05 - 246ms/epoch - 5ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8397 - val_loss: 0.0667 - lr: 1.0000e-05 - 276ms/epoch - 6ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8391 - val_loss: 0.0667 - lr: 1.0000e-05 - 296ms/epoch - 6ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8386 - val_loss: 0.0668 - lr: 1.0000e-05 - 252ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8381 - val_loss: 0.0668 - lr: 1.0000e-05 - 264ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8376 - val_loss: 0.0669 - lr: 1.0000e-05 - 247ms/epoch - 5ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8371 - val_loss: 0.0669 - lr: 1.0000e-05 - 236ms/epoch - 5ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8366 - val_loss: 0.0670 - lr: 1.0000e-05 - 274ms/epoch - 6ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8361 - val_loss: 0.0670 - lr: 1.0000e-05 - 239ms/epoch - 5ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8356 - val_loss: 0.0671 - lr: 1.0000e-05 - 278ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8351 - val_loss: 0.0671 - lr: 1.0000e-05 - 262ms/epoch - 5ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8346 - val_loss: 0.0672 - lr: 1.0000e-05 - 273ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8341 - val_loss: 0.0672 - lr: 1.0000e-05 - 229ms/epoch - 5ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8336 - val_loss: 0.0673 - lr: 1.0000e-05 - 272ms/epoch - 6ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8330 - val_loss: 0.0673 - lr: 1.0000e-05 - 267ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8325 - val_loss: 0.0674 - lr: 1.0000e-05 - 231ms/epoch - 5ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8320 - val_loss: 0.0674 - lr: 1.0000e-05 - 242ms/epoch - 5ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8315 - val_loss: 0.0675 - lr: 1.0000e-05 - 326ms/epoch - 7ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8310 - val_loss: 0.0675 - lr: 1.0000e-05 - 310ms/epoch - 6ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.05005
48/48 - 0s - loss: 0.8305 - val_loss: 0.0676 - lr: 1.0000e-05 - 294ms/epoch - 6ms/step
Epoch 00051: early stopping
SMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 51.87% Accuracy
MSE: 25.081351909450348
RMSE: 5.008128583557969
MAPE: 3.9377037384058786
EMA
EMA([input_arrays], [timeperiod=30])
Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
51
Working on EMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.43 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4231.556, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3761.238, Time=0.05 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.27 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3532.227, Time=0.07 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3394.496, Time=0.09 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=0.90 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.69 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3396.496, Time=0.20 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 2.734 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1693.248
Date: Sun, 12 Dec 2021 AIC 3394.496
Time: 14:40:42 BIC 3413.260
Sample: 0 HQIC 3401.702
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1982 0.003 -389.569 0.000 -1.204 -1.192
ar.L2 -0.8976 0.006 -139.811 0.000 -0.910 -0.885
ar.L3 -0.3984 0.006 -68.662 0.000 -0.410 -0.387
sigma2 3.9230 0.018 215.372 0.000 3.887 3.959
===================================================================================
Ljung-Box (L1) (Q): 14.54 Jarque-Bera (JB): 2462173.05
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.90
Prob(H) (two-sided): 0.00 Kurtosis: 273.82
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.04901, saving model to LSTM4.h5
16/16 - 4s - loss: 1.4775 - val_loss: 0.0490 - lr: 0.0010 - 4s/epoch - 247ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.04901 to 0.04888, saving model to LSTM4.h5
16/16 - 0s - loss: 1.4455 - val_loss: 0.0489 - lr: 0.0010 - 118ms/epoch - 7ms/step
Epoch 3/500
Epoch 00003: val_loss improved from 0.04888 to 0.04882, saving model to LSTM4.h5
16/16 - 0s - loss: 1.4175 - val_loss: 0.0488 - lr: 0.0010 - 113ms/epoch - 7ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.3914 - val_loss: 0.0489 - lr: 0.0010 - 103ms/epoch - 6ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.3659 - val_loss: 0.0492 - lr: 0.0010 - 114ms/epoch - 7ms/step
Epoch 6/500
Epoch 00006: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.3397 - val_loss: 0.0496 - lr: 0.0010 - 102ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00007: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.3125 - val_loss: 0.0501 - lr: 0.0010 - 95ms/epoch - 6ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2943 - val_loss: 0.0502 - lr: 1.0000e-04 - 108ms/epoch - 7ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2915 - val_loss: 0.0503 - lr: 1.0000e-04 - 95ms/epoch - 6ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2889 - val_loss: 0.0503 - lr: 1.0000e-04 - 107ms/epoch - 7ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2862 - val_loss: 0.0504 - lr: 1.0000e-04 - 96ms/epoch - 6ms/step
Epoch 12/500
Epoch 00012: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00012: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2836 - val_loss: 0.0505 - lr: 1.0000e-04 - 93ms/epoch - 6ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2819 - val_loss: 0.0505 - lr: 1.0000e-05 - 98ms/epoch - 6ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2816 - val_loss: 0.0505 - lr: 1.0000e-05 - 105ms/epoch - 7ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2814 - val_loss: 0.0505 - lr: 1.0000e-05 - 98ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2811 - val_loss: 0.0505 - lr: 1.0000e-05 - 91ms/epoch - 6ms/step
Epoch 17/500
Epoch 00017: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00017: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2809 - val_loss: 0.0505 - lr: 1.0000e-05 - 96ms/epoch - 6ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2806 - val_loss: 0.0505 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2804 - val_loss: 0.0505 - lr: 1.0000e-05 - 108ms/epoch - 7ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2801 - val_loss: 0.0505 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2799 - val_loss: 0.0505 - lr: 1.0000e-05 - 110ms/epoch - 7ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2796 - val_loss: 0.0505 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2794 - val_loss: 0.0506 - lr: 1.0000e-05 - 96ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2791 - val_loss: 0.0506 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2789 - val_loss: 0.0506 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2786 - val_loss: 0.0506 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2784 - val_loss: 0.0506 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2781 - val_loss: 0.0506 - lr: 1.0000e-05 - 96ms/epoch - 6ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2779 - val_loss: 0.0506 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2776 - val_loss: 0.0506 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2774 - val_loss: 0.0506 - lr: 1.0000e-05 - 100ms/epoch - 6ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2772 - val_loss: 0.0506 - lr: 1.0000e-05 - 98ms/epoch - 6ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2769 - val_loss: 0.0506 - lr: 1.0000e-05 - 98ms/epoch - 6ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2767 - val_loss: 0.0506 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2764 - val_loss: 0.0506 - lr: 1.0000e-05 - 97ms/epoch - 6ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2762 - val_loss: 0.0507 - lr: 1.0000e-05 - 106ms/epoch - 7ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2759 - val_loss: 0.0507 - lr: 1.0000e-05 - 109ms/epoch - 7ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2757 - val_loss: 0.0507 - lr: 1.0000e-05 - 105ms/epoch - 7ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2754 - val_loss: 0.0507 - lr: 1.0000e-05 - 100ms/epoch - 6ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2752 - val_loss: 0.0507 - lr: 1.0000e-05 - 108ms/epoch - 7ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2749 - val_loss: 0.0507 - lr: 1.0000e-05 - 92ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2747 - val_loss: 0.0507 - lr: 1.0000e-05 - 94ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2744 - val_loss: 0.0507 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2742 - val_loss: 0.0507 - lr: 1.0000e-05 - 112ms/epoch - 7ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2740 - val_loss: 0.0507 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2737 - val_loss: 0.0507 - lr: 1.0000e-05 - 98ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2735 - val_loss: 0.0507 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2732 - val_loss: 0.0507 - lr: 1.0000e-05 - 97ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2730 - val_loss: 0.0508 - lr: 1.0000e-05 - 105ms/epoch - 7ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2727 - val_loss: 0.0508 - lr: 1.0000e-05 - 98ms/epoch - 6ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2725 - val_loss: 0.0508 - lr: 1.0000e-05 - 94ms/epoch - 6ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2722 - val_loss: 0.0508 - lr: 1.0000e-05 - 97ms/epoch - 6ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.04882
16/16 - 0s - loss: 1.2720 - val_loss: 0.0508 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 00053: early stopping
SMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 51.87% Accuracy
MSE: 25.081351909450348
RMSE: 5.008128583557969
MAPE: 3.9377037384058786
EMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 37.86506779592453
RMSE: 6.153459823215273
MAPE: 4.830217084369749
WMA
WMA([input_arrays], [timeperiod=30])
Weighted Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
49
Working on WMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.43 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4264.089, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3793.930, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.26 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3564.923, Time=0.07 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3427.258, Time=0.08 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.25 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.45 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3429.258, Time=0.19 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 2.799 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1709.629
Date: Sun, 12 Dec 2021 AIC 3427.258
Time: 14:42:01 BIC 3446.021
Sample: 0 HQIC 3434.464
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1981 0.003 -389.386 0.000 -1.204 -1.192
ar.L2 -0.8974 0.006 -139.699 0.000 -0.910 -0.885
ar.L3 -0.3983 0.006 -68.737 0.000 -0.410 -0.387
sigma2 4.0860 0.019 215.311 0.000 4.049 4.123
===================================================================================
Ljung-Box (L1) (Q): 14.57 Jarque-Bera (JB): 2460901.70
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.90
Prob(H) (two-sided): 0.00 Kurtosis: 273.75
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.05020, saving model to LSTM4.h5
17/17 - 4s - loss: 1.4028 - val_loss: 0.0502 - lr: 0.0010 - 4s/epoch - 211ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.05020 to 0.04800, saving model to LSTM4.h5
17/17 - 0s - loss: 1.3285 - val_loss: 0.0480 - lr: 0.0010 - 121ms/epoch - 7ms/step
Epoch 3/500
Epoch 00003: val_loss improved from 0.04800 to 0.04527, saving model to LSTM4.h5
17/17 - 0s - loss: 1.2689 - val_loss: 0.0453 - lr: 0.0010 - 122ms/epoch - 7ms/step
Epoch 4/500
Epoch 00004: val_loss improved from 0.04527 to 0.04270, saving model to LSTM4.h5
17/17 - 0s - loss: 1.2122 - val_loss: 0.0427 - lr: 0.0010 - 141ms/epoch - 8ms/step
Epoch 5/500
Epoch 00005: val_loss improved from 0.04270 to 0.04103, saving model to LSTM4.h5
17/17 - 0s - loss: 1.1524 - val_loss: 0.0410 - lr: 0.0010 - 133ms/epoch - 8ms/step
Epoch 6/500
Epoch 00006: val_loss improved from 0.04103 to 0.04060, saving model to LSTM4.h5
17/17 - 0s - loss: 1.0902 - val_loss: 0.0406 - lr: 0.0010 - 138ms/epoch - 8ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.04060
17/17 - 0s - loss: 1.0302 - val_loss: 0.0412 - lr: 0.0010 - 124ms/epoch - 7ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.9764 - val_loss: 0.0424 - lr: 0.0010 - 121ms/epoch - 7ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.9295 - val_loss: 0.0439 - lr: 0.0010 - 108ms/epoch - 6ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8887 - val_loss: 0.0454 - lr: 0.0010 - 120ms/epoch - 7ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00011: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8535 - val_loss: 0.0468 - lr: 0.0010 - 106ms/epoch - 6ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8328 - val_loss: 0.0469 - lr: 1.0000e-04 - 109ms/epoch - 6ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8300 - val_loss: 0.0471 - lr: 1.0000e-04 - 105ms/epoch - 6ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8273 - val_loss: 0.0472 - lr: 1.0000e-04 - 92ms/epoch - 5ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8247 - val_loss: 0.0474 - lr: 1.0000e-04 - 109ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00016: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8221 - val_loss: 0.0475 - lr: 1.0000e-04 - 105ms/epoch - 6ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8204 - val_loss: 0.0475 - lr: 1.0000e-05 - 94ms/epoch - 6ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8201 - val_loss: 0.0475 - lr: 1.0000e-05 - 104ms/epoch - 6ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8199 - val_loss: 0.0476 - lr: 1.0000e-05 - 101ms/epoch - 6ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8196 - val_loss: 0.0476 - lr: 1.0000e-05 - 107ms/epoch - 6ms/step
Epoch 21/500
Epoch 00021: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00021: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8193 - val_loss: 0.0476 - lr: 1.0000e-05 - 101ms/epoch - 6ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8191 - val_loss: 0.0476 - lr: 1.0000e-05 - 104ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8188 - val_loss: 0.0476 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8186 - val_loss: 0.0477 - lr: 1.0000e-05 - 104ms/epoch - 6ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8183 - val_loss: 0.0477 - lr: 1.0000e-05 - 94ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8180 - val_loss: 0.0477 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8178 - val_loss: 0.0477 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8175 - val_loss: 0.0477 - lr: 1.0000e-05 - 113ms/epoch - 7ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8173 - val_loss: 0.0477 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8170 - val_loss: 0.0478 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8167 - val_loss: 0.0478 - lr: 1.0000e-05 - 91ms/epoch - 5ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8165 - val_loss: 0.0478 - lr: 1.0000e-05 - 108ms/epoch - 6ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8162 - val_loss: 0.0478 - lr: 1.0000e-05 - 96ms/epoch - 6ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8159 - val_loss: 0.0478 - lr: 1.0000e-05 - 104ms/epoch - 6ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8157 - val_loss: 0.0479 - lr: 1.0000e-05 - 98ms/epoch - 6ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8154 - val_loss: 0.0479 - lr: 1.0000e-05 - 104ms/epoch - 6ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8151 - val_loss: 0.0479 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8149 - val_loss: 0.0479 - lr: 1.0000e-05 - 114ms/epoch - 7ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8146 - val_loss: 0.0480 - lr: 1.0000e-05 - 100ms/epoch - 6ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8143 - val_loss: 0.0480 - lr: 1.0000e-05 - 101ms/epoch - 6ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8141 - val_loss: 0.0480 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8138 - val_loss: 0.0480 - lr: 1.0000e-05 - 120ms/epoch - 7ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8135 - val_loss: 0.0480 - lr: 1.0000e-05 - 134ms/epoch - 8ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8133 - val_loss: 0.0481 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8130 - val_loss: 0.0481 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8127 - val_loss: 0.0481 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8125 - val_loss: 0.0481 - lr: 1.0000e-05 - 113ms/epoch - 7ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8122 - val_loss: 0.0482 - lr: 1.0000e-05 - 106ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8119 - val_loss: 0.0482 - lr: 1.0000e-05 - 101ms/epoch - 6ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8117 - val_loss: 0.0482 - lr: 1.0000e-05 - 114ms/epoch - 7ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8114 - val_loss: 0.0482 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8111 - val_loss: 0.0482 - lr: 1.0000e-05 - 108ms/epoch - 6ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8108 - val_loss: 0.0483 - lr: 1.0000e-05 - 122ms/epoch - 7ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8106 - val_loss: 0.0483 - lr: 1.0000e-05 - 104ms/epoch - 6ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8103 - val_loss: 0.0483 - lr: 1.0000e-05 - 113ms/epoch - 7ms/step
Epoch 56/500
Epoch 00056: val_loss did not improve from 0.04060
17/17 - 0s - loss: 0.8100 - val_loss: 0.0483 - lr: 1.0000e-05 - 125ms/epoch - 7ms/step
Epoch 00056: early stopping
SMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 51.87% Accuracy
MSE: 25.081351909450348
RMSE: 5.008128583557969
MAPE: 3.9377037384058786
EMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 37.86506779592453
RMSE: 6.153459823215273
MAPE: 4.830217084369749
WMA
Prediction vs Close: 53.36% Accuracy
Prediction vs Prediction: 47.76% Accuracy
MSE: 53.37812690087662
RMSE: 7.306033595657539
MAPE: 5.95316326588041
DEMA
DEMA([input_arrays], [timeperiod=30])
Double Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
89
Working on DEMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.42 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4436.126, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3965.317, Time=0.05 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.40 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3736.589, Time=0.06 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3598.951, Time=0.09 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.00 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.98 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3600.951, Time=0.17 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.209 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1795.475
Date: Sun, 12 Dec 2021 AIC 3598.951
Time: 14:43:22 BIC 3617.714
Sample: 0 HQIC 3606.157
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1983 0.003 -389.581 0.000 -1.204 -1.192
ar.L2 -0.8973 0.006 -139.732 0.000 -0.910 -0.885
ar.L3 -0.3983 0.006 -68.649 0.000 -0.410 -0.387
sigma2 5.0573 0.023 215.292 0.000 5.011 5.103
===================================================================================
Ljung-Box (L1) (Q): 14.41 Jarque-Bera (JB): 2460553.80
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.89
Prob(H) (two-sided): 0.00 Kurtosis: 273.74
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.05015, saving model to LSTM4.h5
10/10 - 4s - loss: 1.4573 - val_loss: 0.0502 - lr: 0.0010 - 4s/epoch - 359ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.05015 to 0.05012, saving model to LSTM4.h5
10/10 - 0s - loss: 1.4383 - val_loss: 0.0501 - lr: 0.0010 - 105ms/epoch - 10ms/step
Epoch 3/500
Epoch 00003: val_loss improved from 0.05012 to 0.05001, saving model to LSTM4.h5
10/10 - 0s - loss: 1.4219 - val_loss: 0.0500 - lr: 0.0010 - 93ms/epoch - 9ms/step
Epoch 4/500
Epoch 00004: val_loss improved from 0.05001 to 0.04985, saving model to LSTM4.h5
10/10 - 0s - loss: 1.4070 - val_loss: 0.0499 - lr: 0.0010 - 83ms/epoch - 8ms/step
Epoch 5/500
Epoch 00005: val_loss improved from 0.04985 to 0.04968, saving model to LSTM4.h5
10/10 - 0s - loss: 1.3928 - val_loss: 0.0497 - lr: 0.0010 - 79ms/epoch - 8ms/step
Epoch 6/500
Epoch 00006: val_loss improved from 0.04968 to 0.04952, saving model to LSTM4.h5
10/10 - 0s - loss: 1.3788 - val_loss: 0.0495 - lr: 0.0010 - 96ms/epoch - 10ms/step
Epoch 7/500
Epoch 00007: val_loss improved from 0.04952 to 0.04937, saving model to LSTM4.h5
10/10 - 0s - loss: 1.3643 - val_loss: 0.0494 - lr: 0.0010 - 106ms/epoch - 11ms/step
Epoch 8/500
Epoch 00008: val_loss improved from 0.04937 to 0.04925, saving model to LSTM4.h5
10/10 - 0s - loss: 1.3487 - val_loss: 0.0492 - lr: 0.0010 - 97ms/epoch - 10ms/step
Epoch 9/500
Epoch 00009: val_loss improved from 0.04925 to 0.04916, saving model to LSTM4.h5
10/10 - 0s - loss: 1.3317 - val_loss: 0.0492 - lr: 0.0010 - 101ms/epoch - 10ms/step
Epoch 10/500
Epoch 00010: val_loss improved from 0.04916 to 0.04913, saving model to LSTM4.h5
10/10 - 0s - loss: 1.3129 - val_loss: 0.0491 - lr: 0.0010 - 94ms/epoch - 9ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.2920 - val_loss: 0.0492 - lr: 0.0010 - 81ms/epoch - 8ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.2690 - val_loss: 0.0493 - lr: 0.0010 - 72ms/epoch - 7ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.2440 - val_loss: 0.0495 - lr: 0.0010 - 65ms/epoch - 7ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.2171 - val_loss: 0.0498 - lr: 0.0010 - 78ms/epoch - 8ms/step
Epoch 15/500
Epoch 00015: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00015: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1885 - val_loss: 0.0504 - lr: 0.0010 - 63ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1676 - val_loss: 0.0504 - lr: 1.0000e-04 - 68ms/epoch - 7ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1646 - val_loss: 0.0505 - lr: 1.0000e-04 - 65ms/epoch - 7ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1617 - val_loss: 0.0506 - lr: 1.0000e-04 - 68ms/epoch - 7ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1589 - val_loss: 0.0507 - lr: 1.0000e-04 - 75ms/epoch - 8ms/step
Epoch 20/500
Epoch 00020: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00020: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1562 - val_loss: 0.0507 - lr: 1.0000e-04 - 83ms/epoch - 8ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1543 - val_loss: 0.0507 - lr: 1.0000e-05 - 86ms/epoch - 9ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1540 - val_loss: 0.0508 - lr: 1.0000e-05 - 79ms/epoch - 8ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1537 - val_loss: 0.0508 - lr: 1.0000e-05 - 83ms/epoch - 8ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1535 - val_loss: 0.0508 - lr: 1.0000e-05 - 69ms/epoch - 7ms/step
Epoch 25/500
Epoch 00025: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00025: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1532 - val_loss: 0.0508 - lr: 1.0000e-05 - 79ms/epoch - 8ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1530 - val_loss: 0.0508 - lr: 1.0000e-05 - 80ms/epoch - 8ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1527 - val_loss: 0.0508 - lr: 1.0000e-05 - 66ms/epoch - 7ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1525 - val_loss: 0.0508 - lr: 1.0000e-05 - 66ms/epoch - 7ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1522 - val_loss: 0.0508 - lr: 1.0000e-05 - 70ms/epoch - 7ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1520 - val_loss: 0.0508 - lr: 1.0000e-05 - 69ms/epoch - 7ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1517 - val_loss: 0.0508 - lr: 1.0000e-05 - 73ms/epoch - 7ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1515 - val_loss: 0.0508 - lr: 1.0000e-05 - 74ms/epoch - 7ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1512 - val_loss: 0.0509 - lr: 1.0000e-05 - 89ms/epoch - 9ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1510 - val_loss: 0.0509 - lr: 1.0000e-05 - 82ms/epoch - 8ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1507 - val_loss: 0.0509 - lr: 1.0000e-05 - 68ms/epoch - 7ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1505 - val_loss: 0.0509 - lr: 1.0000e-05 - 75ms/epoch - 7ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1503 - val_loss: 0.0509 - lr: 1.0000e-05 - 78ms/epoch - 8ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1500 - val_loss: 0.0509 - lr: 1.0000e-05 - 72ms/epoch - 7ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1498 - val_loss: 0.0509 - lr: 1.0000e-05 - 82ms/epoch - 8ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1495 - val_loss: 0.0509 - lr: 1.0000e-05 - 68ms/epoch - 7ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1493 - val_loss: 0.0509 - lr: 1.0000e-05 - 77ms/epoch - 8ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1491 - val_loss: 0.0509 - lr: 1.0000e-05 - 65ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1488 - val_loss: 0.0509 - lr: 1.0000e-05 - 77ms/epoch - 8ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1486 - val_loss: 0.0510 - lr: 1.0000e-05 - 83ms/epoch - 8ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1484 - val_loss: 0.0510 - lr: 1.0000e-05 - 81ms/epoch - 8ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1481 - val_loss: 0.0510 - lr: 1.0000e-05 - 82ms/epoch - 8ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1479 - val_loss: 0.0510 - lr: 1.0000e-05 - 74ms/epoch - 7ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1476 - val_loss: 0.0510 - lr: 1.0000e-05 - 72ms/epoch - 7ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1474 - val_loss: 0.0510 - lr: 1.0000e-05 - 70ms/epoch - 7ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1472 - val_loss: 0.0510 - lr: 1.0000e-05 - 67ms/epoch - 7ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1469 - val_loss: 0.0510 - lr: 1.0000e-05 - 68ms/epoch - 7ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1467 - val_loss: 0.0510 - lr: 1.0000e-05 - 75ms/epoch - 7ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1465 - val_loss: 0.0510 - lr: 1.0000e-05 - 79ms/epoch - 8ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1462 - val_loss: 0.0511 - lr: 1.0000e-05 - 80ms/epoch - 8ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1460 - val_loss: 0.0511 - lr: 1.0000e-05 - 67ms/epoch - 7ms/step
Epoch 56/500
Epoch 00056: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1458 - val_loss: 0.0511 - lr: 1.0000e-05 - 89ms/epoch - 9ms/step
Epoch 57/500
Epoch 00057: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1455 - val_loss: 0.0511 - lr: 1.0000e-05 - 83ms/epoch - 8ms/step
Epoch 58/500
Epoch 00058: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1453 - val_loss: 0.0511 - lr: 1.0000e-05 - 74ms/epoch - 7ms/step
Epoch 59/500
Epoch 00059: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1451 - val_loss: 0.0511 - lr: 1.0000e-05 - 70ms/epoch - 7ms/step
Epoch 60/500
Epoch 00060: val_loss did not improve from 0.04913
10/10 - 0s - loss: 1.1448 - val_loss: 0.0511 - lr: 1.0000e-05 - 71ms/epoch - 7ms/step
Epoch 00060: early stopping
SMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 51.87% Accuracy
MSE: 25.081351909450348
RMSE: 5.008128583557969
MAPE: 3.9377037384058786
EMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 37.86506779592453
RMSE: 6.153459823215273
MAPE: 4.830217084369749
WMA
Prediction vs Close: 53.36% Accuracy
Prediction vs Prediction: 47.76% Accuracy
MSE: 53.37812690087662
RMSE: 7.306033595657539
MAPE: 5.95316326588041
DEMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 44.78% Accuracy
MSE: 133.19678136227444
RMSE: 11.541090995320783
MAPE: 10.29859546777107
KAMA
KAMA([input_arrays], [timeperiod=30])
Kaufman Adaptive Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
18
Working on KAMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.37 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4190.464, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3724.371, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.28 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3494.154, Time=0.07 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3357.435, Time=0.09 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.21 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.76 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3359.435, Time=0.21 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.077 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1674.717
Date: Sun, 12 Dec 2021 AIC 3357.435
Time: 14:44:41 BIC 3376.198
Sample: 0 HQIC 3364.641
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1955 0.003 -381.246 0.000 -1.202 -1.189
ar.L2 -0.8964 0.007 -135.835 0.000 -0.909 -0.883
ar.L3 -0.3971 0.006 -67.229 0.000 -0.409 -0.385
sigma2 3.7466 0.018 211.623 0.000 3.712 3.781
===================================================================================
Ljung-Box (L1) (Q): 14.20 Jarque-Bera (JB): 2338363.32
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.01 Skew: 3.76
Prob(H) (two-sided): 0.00 Kurtosis: 266.93
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.05434, saving model to LSTM4.h5
45/45 - 4s - loss: 1.4199 - val_loss: 0.0543 - lr: 0.0010 - 4s/epoch - 97ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.05434
45/45 - 0s - loss: 1.3785 - val_loss: 0.0549 - lr: 0.0010 - 231ms/epoch - 5ms/step
Epoch 3/500
Epoch 00003: val_loss improved from 0.05434 to 0.05416, saving model to LSTM4.h5
45/45 - 0s - loss: 1.3278 - val_loss: 0.0542 - lr: 0.0010 - 270ms/epoch - 6ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.05416
45/45 - 0s - loss: 1.2376 - val_loss: 0.0562 - lr: 0.0010 - 237ms/epoch - 5ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.05416
45/45 - 0s - loss: 1.1316 - val_loss: 0.0603 - lr: 0.0010 - 252ms/epoch - 6ms/step
Epoch 6/500
Epoch 00006: val_loss did not improve from 0.05416
45/45 - 0s - loss: 1.0605 - val_loss: 0.0647 - lr: 0.0010 - 239ms/epoch - 5ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.05416
45/45 - 0s - loss: 1.0065 - val_loss: 0.0694 - lr: 0.0010 - 267ms/epoch - 6ms/step
Epoch 8/500
Epoch 00008: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00008: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9623 - val_loss: 0.0744 - lr: 0.0010 - 249ms/epoch - 6ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9381 - val_loss: 0.0749 - lr: 1.0000e-04 - 232ms/epoch - 5ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9346 - val_loss: 0.0754 - lr: 1.0000e-04 - 268ms/epoch - 6ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9311 - val_loss: 0.0759 - lr: 1.0000e-04 - 258ms/epoch - 6ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9276 - val_loss: 0.0765 - lr: 1.0000e-04 - 230ms/epoch - 5ms/step
Epoch 13/500
Epoch 00013: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00013: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9242 - val_loss: 0.0770 - lr: 1.0000e-04 - 230ms/epoch - 5ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9221 - val_loss: 0.0771 - lr: 1.0000e-05 - 228ms/epoch - 5ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9218 - val_loss: 0.0771 - lr: 1.0000e-05 - 238ms/epoch - 5ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9215 - val_loss: 0.0772 - lr: 1.0000e-05 - 222ms/epoch - 5ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9211 - val_loss: 0.0772 - lr: 1.0000e-05 - 283ms/epoch - 6ms/step
Epoch 18/500
Epoch 00018: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00018: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9208 - val_loss: 0.0773 - lr: 1.0000e-05 - 270ms/epoch - 6ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9204 - val_loss: 0.0774 - lr: 1.0000e-05 - 308ms/epoch - 7ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9201 - val_loss: 0.0774 - lr: 1.0000e-05 - 242ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9197 - val_loss: 0.0775 - lr: 1.0000e-05 - 224ms/epoch - 5ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9194 - val_loss: 0.0776 - lr: 1.0000e-05 - 268ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9191 - val_loss: 0.0776 - lr: 1.0000e-05 - 246ms/epoch - 5ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9187 - val_loss: 0.0777 - lr: 1.0000e-05 - 265ms/epoch - 6ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9184 - val_loss: 0.0778 - lr: 1.0000e-05 - 230ms/epoch - 5ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9180 - val_loss: 0.0778 - lr: 1.0000e-05 - 289ms/epoch - 6ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9177 - val_loss: 0.0779 - lr: 1.0000e-05 - 256ms/epoch - 6ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9173 - val_loss: 0.0780 - lr: 1.0000e-05 - 228ms/epoch - 5ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9170 - val_loss: 0.0780 - lr: 1.0000e-05 - 243ms/epoch - 5ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9166 - val_loss: 0.0781 - lr: 1.0000e-05 - 243ms/epoch - 5ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9163 - val_loss: 0.0782 - lr: 1.0000e-05 - 274ms/epoch - 6ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9159 - val_loss: 0.0783 - lr: 1.0000e-05 - 248ms/epoch - 6ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9156 - val_loss: 0.0783 - lr: 1.0000e-05 - 303ms/epoch - 7ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9152 - val_loss: 0.0784 - lr: 1.0000e-05 - 239ms/epoch - 5ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9149 - val_loss: 0.0785 - lr: 1.0000e-05 - 238ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9145 - val_loss: 0.0786 - lr: 1.0000e-05 - 246ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9142 - val_loss: 0.0786 - lr: 1.0000e-05 - 246ms/epoch - 5ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9138 - val_loss: 0.0787 - lr: 1.0000e-05 - 235ms/epoch - 5ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9135 - val_loss: 0.0788 - lr: 1.0000e-05 - 266ms/epoch - 6ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9131 - val_loss: 0.0789 - lr: 1.0000e-05 - 283ms/epoch - 6ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9128 - val_loss: 0.0790 - lr: 1.0000e-05 - 282ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9124 - val_loss: 0.0790 - lr: 1.0000e-05 - 250ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9121 - val_loss: 0.0791 - lr: 1.0000e-05 - 256ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9117 - val_loss: 0.0792 - lr: 1.0000e-05 - 250ms/epoch - 6ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9114 - val_loss: 0.0793 - lr: 1.0000e-05 - 265ms/epoch - 6ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9110 - val_loss: 0.0794 - lr: 1.0000e-05 - 295ms/epoch - 7ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9107 - val_loss: 0.0794 - lr: 1.0000e-05 - 251ms/epoch - 6ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9103 - val_loss: 0.0795 - lr: 1.0000e-05 - 248ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9100 - val_loss: 0.0796 - lr: 1.0000e-05 - 284ms/epoch - 6ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9096 - val_loss: 0.0797 - lr: 1.0000e-05 - 267ms/epoch - 6ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9093 - val_loss: 0.0798 - lr: 1.0000e-05 - 224ms/epoch - 5ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9089 - val_loss: 0.0799 - lr: 1.0000e-05 - 252ms/epoch - 6ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.05416
45/45 - 0s - loss: 0.9086 - val_loss: 0.0800 - lr: 1.0000e-05 - 275ms/epoch - 6ms/step
Epoch 00053: early stopping
SMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 51.87% Accuracy
MSE: 25.081351909450348
RMSE: 5.008128583557969
MAPE: 3.9377037384058786
EMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 37.86506779592453
RMSE: 6.153459823215273
MAPE: 4.830217084369749
WMA
Prediction vs Close: 53.36% Accuracy
Prediction vs Prediction: 47.76% Accuracy
MSE: 53.37812690087662
RMSE: 7.306033595657539
MAPE: 5.95316326588041
DEMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 44.78% Accuracy
MSE: 133.19678136227444
RMSE: 11.541090995320783
MAPE: 10.29859546777107
KAMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 50.37% Accuracy
MSE: 20.693935177088164
RMSE: 4.549058713304123
MAPE: 3.6577262429810227
MIDPOINT
MIDPOINT([input_arrays], [timeperiod=14])
MidPoint over period (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 14
Outputs:
real
14
Working on MIDPOINT predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.36 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4212.289, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3747.746, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.24 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3523.401, Time=0.08 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3387.759, Time=0.11 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.31 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.93 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3389.758, Time=0.22 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.314 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1689.879
Date: Sun, 12 Dec 2021 AIC 3387.759
Time: 14:46:10 BIC 3406.522
Sample: 0 HQIC 3394.964
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1878 0.003 -345.315 0.000 -1.195 -1.181
ar.L2 -0.8876 0.007 -121.809 0.000 -0.902 -0.873
ar.L3 -0.3957 0.007 -60.127 0.000 -0.409 -0.383
sigma2 3.8904 0.020 193.404 0.000 3.851 3.930
===================================================================================
Ljung-Box (L1) (Q): 13.21 Jarque-Bera (JB): 1659080.01
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.08 Skew: 3.28
Prob(H) (two-sided): 0.00 Kurtosis: 225.31
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.05739, saving model to LSTM4.h5
58/58 - 4s - loss: 1.3554 - val_loss: 0.0574 - lr: 0.0010 - 4s/epoch - 73ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.05739 to 0.05277, saving model to LSTM4.h5
58/58 - 0s - loss: 1.1751 - val_loss: 0.0528 - lr: 0.0010 - 331ms/epoch - 6ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.05277
58/58 - 0s - loss: 1.0556 - val_loss: 0.0541 - lr: 0.0010 - 329ms/epoch - 6ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.9570 - val_loss: 0.0582 - lr: 0.0010 - 287ms/epoch - 5ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.8763 - val_loss: 0.0633 - lr: 0.0010 - 291ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.8146 - val_loss: 0.0687 - lr: 0.0010 - 305ms/epoch - 5ms/step
Epoch 7/500
Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00007: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7700 - val_loss: 0.0739 - lr: 0.0010 - 303ms/epoch - 5ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7483 - val_loss: 0.0744 - lr: 1.0000e-04 - 297ms/epoch - 5ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7453 - val_loss: 0.0750 - lr: 1.0000e-04 - 326ms/epoch - 6ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7422 - val_loss: 0.0756 - lr: 1.0000e-04 - 293ms/epoch - 5ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7392 - val_loss: 0.0762 - lr: 1.0000e-04 - 289ms/epoch - 5ms/step
Epoch 12/500
Epoch 00012: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00012: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7363 - val_loss: 0.0768 - lr: 1.0000e-04 - 314ms/epoch - 5ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7344 - val_loss: 0.0768 - lr: 1.0000e-05 - 300ms/epoch - 5ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7341 - val_loss: 0.0769 - lr: 1.0000e-05 - 312ms/epoch - 5ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7338 - val_loss: 0.0770 - lr: 1.0000e-05 - 331ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7335 - val_loss: 0.0770 - lr: 1.0000e-05 - 331ms/epoch - 6ms/step
Epoch 17/500
Epoch 00017: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00017: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7332 - val_loss: 0.0771 - lr: 1.0000e-05 - 310ms/epoch - 5ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7329 - val_loss: 0.0772 - lr: 1.0000e-05 - 300ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7326 - val_loss: 0.0773 - lr: 1.0000e-05 - 306ms/epoch - 5ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7323 - val_loss: 0.0773 - lr: 1.0000e-05 - 298ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7320 - val_loss: 0.0774 - lr: 1.0000e-05 - 364ms/epoch - 6ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7316 - val_loss: 0.0775 - lr: 1.0000e-05 - 358ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7313 - val_loss: 0.0776 - lr: 1.0000e-05 - 286ms/epoch - 5ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7310 - val_loss: 0.0777 - lr: 1.0000e-05 - 295ms/epoch - 5ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7306 - val_loss: 0.0778 - lr: 1.0000e-05 - 324ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7303 - val_loss: 0.0778 - lr: 1.0000e-05 - 291ms/epoch - 5ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7300 - val_loss: 0.0779 - lr: 1.0000e-05 - 310ms/epoch - 5ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7296 - val_loss: 0.0780 - lr: 1.0000e-05 - 346ms/epoch - 6ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7293 - val_loss: 0.0781 - lr: 1.0000e-05 - 331ms/epoch - 6ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7290 - val_loss: 0.0782 - lr: 1.0000e-05 - 315ms/epoch - 5ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7286 - val_loss: 0.0783 - lr: 1.0000e-05 - 335ms/epoch - 6ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7283 - val_loss: 0.0784 - lr: 1.0000e-05 - 293ms/epoch - 5ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7279 - val_loss: 0.0785 - lr: 1.0000e-05 - 311ms/epoch - 5ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7276 - val_loss: 0.0786 - lr: 1.0000e-05 - 326ms/epoch - 6ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7272 - val_loss: 0.0787 - lr: 1.0000e-05 - 316ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7269 - val_loss: 0.0788 - lr: 1.0000e-05 - 310ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7265 - val_loss: 0.0789 - lr: 1.0000e-05 - 342ms/epoch - 6ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7262 - val_loss: 0.0790 - lr: 1.0000e-05 - 331ms/epoch - 6ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7258 - val_loss: 0.0791 - lr: 1.0000e-05 - 294ms/epoch - 5ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7255 - val_loss: 0.0792 - lr: 1.0000e-05 - 304ms/epoch - 5ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7252 - val_loss: 0.0793 - lr: 1.0000e-05 - 315ms/epoch - 5ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7248 - val_loss: 0.0794 - lr: 1.0000e-05 - 285ms/epoch - 5ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7245 - val_loss: 0.0796 - lr: 1.0000e-05 - 301ms/epoch - 5ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7241 - val_loss: 0.0797 - lr: 1.0000e-05 - 302ms/epoch - 5ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7238 - val_loss: 0.0798 - lr: 1.0000e-05 - 305ms/epoch - 5ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7234 - val_loss: 0.0799 - lr: 1.0000e-05 - 314ms/epoch - 5ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7231 - val_loss: 0.0800 - lr: 1.0000e-05 - 333ms/epoch - 6ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7227 - val_loss: 0.0801 - lr: 1.0000e-05 - 325ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7224 - val_loss: 0.0802 - lr: 1.0000e-05 - 318ms/epoch - 5ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7220 - val_loss: 0.0804 - lr: 1.0000e-05 - 304ms/epoch - 5ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7217 - val_loss: 0.0805 - lr: 1.0000e-05 - 301ms/epoch - 5ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.05277
58/58 - 0s - loss: 0.7213 - val_loss: 0.0806 - lr: 1.0000e-05 - 308ms/epoch - 5ms/step
Epoch 00052: early stopping
SMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 51.87% Accuracy
MSE: 25.081351909450348
RMSE: 5.008128583557969
MAPE: 3.9377037384058786
EMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 37.86506779592453
RMSE: 6.153459823215273
MAPE: 4.830217084369749
WMA
Prediction vs Close: 53.36% Accuracy
Prediction vs Prediction: 47.76% Accuracy
MSE: 53.37812690087662
RMSE: 7.306033595657539
MAPE: 5.95316326588041
DEMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 44.78% Accuracy
MSE: 133.19678136227444
RMSE: 11.541090995320783
MAPE: 10.29859546777107
KAMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 50.37% Accuracy
MSE: 20.693935177088164
RMSE: 4.549058713304123
MAPE: 3.6577262429810227
MIDPOINT
Prediction vs Close: 48.13% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 18.24421544500263
RMSE: 4.271324788049093
MAPE: 3.3887721441386436
T3
T3([input_arrays], [timeperiod=5], [vfactor=0.7])
Triple Exponential Moving Average (T3) (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 5
vfactor: 0.7
Outputs:
real
19
Working on T3 predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.41 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4414.515, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3944.062, Time=0.04 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.36 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3715.173, Time=0.07 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3577.471, Time=0.09 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.47 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.58 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3579.471, Time=0.20 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.258 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1784.736
Date: Sun, 12 Dec 2021 AIC 3577.471
Time: 14:47:45 BIC 3596.235
Sample: 0 HQIC 3584.677
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1982 0.003 -389.844 0.000 -1.204 -1.192
ar.L2 -0.8974 0.006 -139.861 0.000 -0.910 -0.885
ar.L3 -0.3983 0.006 -68.862 0.000 -0.410 -0.387
sigma2 4.9242 0.023 215.469 0.000 4.879 4.969
===================================================================================
Ljung-Box (L1) (Q): 14.55 Jarque-Bera (JB): 2468024.38
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 3.90
Prob(H) (two-sided): 0.00 Kurtosis: 274.15
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.04850, saving model to LSTM4.h5
43/43 - 4s - loss: 1.4060 - val_loss: 0.0485 - lr: 0.0010 - 4s/epoch - 88ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.3489 - val_loss: 0.0493 - lr: 0.0010 - 247ms/epoch - 6ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.2894 - val_loss: 0.0512 - lr: 0.0010 - 209ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.2311 - val_loss: 0.0529 - lr: 0.0010 - 228ms/epoch - 5ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.1726 - val_loss: 0.0550 - lr: 0.0010 - 242ms/epoch - 6ms/step
Epoch 6/500
Epoch 00006: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00006: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.1181 - val_loss: 0.0574 - lr: 0.0010 - 234ms/epoch - 5ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0859 - val_loss: 0.0576 - lr: 1.0000e-04 - 219ms/epoch - 5ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0812 - val_loss: 0.0579 - lr: 1.0000e-04 - 226ms/epoch - 5ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0765 - val_loss: 0.0582 - lr: 1.0000e-04 - 239ms/epoch - 6ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0720 - val_loss: 0.0585 - lr: 1.0000e-04 - 263ms/epoch - 6ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00011: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0675 - val_loss: 0.0588 - lr: 1.0000e-04 - 248ms/epoch - 6ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0647 - val_loss: 0.0588 - lr: 1.0000e-05 - 246ms/epoch - 6ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0643 - val_loss: 0.0589 - lr: 1.0000e-05 - 227ms/epoch - 5ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0638 - val_loss: 0.0589 - lr: 1.0000e-05 - 240ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0634 - val_loss: 0.0589 - lr: 1.0000e-05 - 250ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00016: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0630 - val_loss: 0.0590 - lr: 1.0000e-05 - 244ms/epoch - 6ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0625 - val_loss: 0.0590 - lr: 1.0000e-05 - 283ms/epoch - 7ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0621 - val_loss: 0.0590 - lr: 1.0000e-05 - 235ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0616 - val_loss: 0.0591 - lr: 1.0000e-05 - 209ms/epoch - 5ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0612 - val_loss: 0.0591 - lr: 1.0000e-05 - 218ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0608 - val_loss: 0.0591 - lr: 1.0000e-05 - 226ms/epoch - 5ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0603 - val_loss: 0.0592 - lr: 1.0000e-05 - 250ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0599 - val_loss: 0.0592 - lr: 1.0000e-05 - 285ms/epoch - 7ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0595 - val_loss: 0.0592 - lr: 1.0000e-05 - 225ms/epoch - 5ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0590 - val_loss: 0.0593 - lr: 1.0000e-05 - 277ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0586 - val_loss: 0.0593 - lr: 1.0000e-05 - 288ms/epoch - 7ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0582 - val_loss: 0.0593 - lr: 1.0000e-05 - 223ms/epoch - 5ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0577 - val_loss: 0.0594 - lr: 1.0000e-05 - 240ms/epoch - 6ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0573 - val_loss: 0.0594 - lr: 1.0000e-05 - 276ms/epoch - 6ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0568 - val_loss: 0.0595 - lr: 1.0000e-05 - 252ms/epoch - 6ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0564 - val_loss: 0.0595 - lr: 1.0000e-05 - 218ms/epoch - 5ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0560 - val_loss: 0.0595 - lr: 1.0000e-05 - 216ms/epoch - 5ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0555 - val_loss: 0.0596 - lr: 1.0000e-05 - 237ms/epoch - 6ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0551 - val_loss: 0.0596 - lr: 1.0000e-05 - 249ms/epoch - 6ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0547 - val_loss: 0.0596 - lr: 1.0000e-05 - 216ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0542 - val_loss: 0.0597 - lr: 1.0000e-05 - 216ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0538 - val_loss: 0.0597 - lr: 1.0000e-05 - 225ms/epoch - 5ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0534 - val_loss: 0.0598 - lr: 1.0000e-05 - 237ms/epoch - 6ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0529 - val_loss: 0.0598 - lr: 1.0000e-05 - 269ms/epoch - 6ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0525 - val_loss: 0.0598 - lr: 1.0000e-05 - 241ms/epoch - 6ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0521 - val_loss: 0.0599 - lr: 1.0000e-05 - 247ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0516 - val_loss: 0.0599 - lr: 1.0000e-05 - 242ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0512 - val_loss: 0.0599 - lr: 1.0000e-05 - 248ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0508 - val_loss: 0.0600 - lr: 1.0000e-05 - 220ms/epoch - 5ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0503 - val_loss: 0.0600 - lr: 1.0000e-05 - 228ms/epoch - 5ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0499 - val_loss: 0.0601 - lr: 1.0000e-05 - 249ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0495 - val_loss: 0.0601 - lr: 1.0000e-05 - 254ms/epoch - 6ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0490 - val_loss: 0.0601 - lr: 1.0000e-05 - 242ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0486 - val_loss: 0.0602 - lr: 1.0000e-05 - 217ms/epoch - 5ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0482 - val_loss: 0.0602 - lr: 1.0000e-05 - 240ms/epoch - 6ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.04850
43/43 - 0s - loss: 1.0478 - val_loss: 0.0603 - lr: 1.0000e-05 - 222ms/epoch - 5ms/step
Epoch 00051: early stopping
SMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 51.87% Accuracy
MSE: 25.081351909450348
RMSE: 5.008128583557969
MAPE: 3.9377037384058786
EMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 37.86506779592453
RMSE: 6.153459823215273
MAPE: 4.830217084369749
WMA
Prediction vs Close: 53.36% Accuracy
Prediction vs Prediction: 47.76% Accuracy
MSE: 53.37812690087662
RMSE: 7.306033595657539
MAPE: 5.95316326588041
DEMA
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 44.78% Accuracy
MSE: 133.19678136227444
RMSE: 11.541090995320783
MAPE: 10.29859546777107
KAMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 50.37% Accuracy
MSE: 20.693935177088164
RMSE: 4.549058713304123
MAPE: 3.6577262429810227
MIDPOINT
Prediction vs Close: 48.13% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 18.24421544500263
RMSE: 4.271324788049093
MAPE: 3.3887721441386436
T3
Prediction vs Close: 53.73% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 74.9216743993189
RMSE: 8.655730725901707
MAPE: 7.03412901576443
TEMA
TEMA([input_arrays], [timeperiod=30])
Triple Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
9
Working on TEMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=inf, Time=0.45 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=4352.703, Time=0.03 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=3889.412, Time=0.05 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=inf, Time=0.28 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=3689.930, Time=0.06 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=3574.245, Time=0.08 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=inf, Time=1.23 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=inf, Time=0.80 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=3576.245, Time=0.18 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 3.171 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood -1783.123
Date: Sun, 12 Dec 2021 AIC 3574.245
Time: 14:49:16 BIC 3593.008
Sample: 0 HQIC 3581.451
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
ar.L1 -1.1480 0.004 -302.430 0.000 -1.155 -1.141
ar.L2 -0.8300 0.008 -99.682 0.000 -0.846 -0.814
ar.L3 -0.3687 0.007 -50.527 0.000 -0.383 -0.354
sigma2 4.9055 0.028 175.970 0.000 4.851 4.960
===================================================================================
Ljung-Box (L1) (Q): 11.61 Jarque-Bera (JB): 1261976.58
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.16 Skew: 2.52
Prob(H) (two-sided): 0.00 Kurtosis: 196.90
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.04545, saving model to LSTM4.h5
90/90 - 4s - loss: 1.3805 - val_loss: 0.0455 - lr: 0.0010 - 4s/epoch - 44ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.04545
90/90 - 0s - loss: 1.2203 - val_loss: 0.0467 - lr: 0.0010 - 476ms/epoch - 5ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.04545
90/90 - 0s - loss: 1.0618 - val_loss: 0.0520 - lr: 0.0010 - 457ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.9221 - val_loss: 0.0615 - lr: 0.0010 - 446ms/epoch - 5ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.8392 - val_loss: 0.0696 - lr: 0.0010 - 448ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00006: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7895 - val_loss: 0.0768 - lr: 0.0010 - 464ms/epoch - 5ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.04545
90/90 - 1s - loss: 0.7661 - val_loss: 0.0775 - lr: 1.0000e-04 - 532ms/epoch - 6ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7627 - val_loss: 0.0782 - lr: 1.0000e-04 - 447ms/epoch - 5ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.04545
90/90 - 1s - loss: 0.7592 - val_loss: 0.0790 - lr: 1.0000e-04 - 526ms/epoch - 6ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7558 - val_loss: 0.0798 - lr: 1.0000e-04 - 463ms/epoch - 5ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00011: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7524 - val_loss: 0.0807 - lr: 1.0000e-04 - 447ms/epoch - 5ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.04545
90/90 - 1s - loss: 0.7503 - val_loss: 0.0807 - lr: 1.0000e-05 - 530ms/epoch - 6ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7499 - val_loss: 0.0808 - lr: 1.0000e-05 - 421ms/epoch - 5ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.04545
90/90 - 1s - loss: 0.7496 - val_loss: 0.0809 - lr: 1.0000e-05 - 539ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.04545
90/90 - 1s - loss: 0.7492 - val_loss: 0.0810 - lr: 1.0000e-05 - 513ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00016: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7488 - val_loss: 0.0811 - lr: 1.0000e-05 - 462ms/epoch - 5ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7485 - val_loss: 0.0812 - lr: 1.0000e-05 - 459ms/epoch - 5ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7481 - val_loss: 0.0813 - lr: 1.0000e-05 - 457ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7477 - val_loss: 0.0815 - lr: 1.0000e-05 - 435ms/epoch - 5ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7473 - val_loss: 0.0816 - lr: 1.0000e-05 - 459ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7470 - val_loss: 0.0817 - lr: 1.0000e-05 - 434ms/epoch - 5ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7466 - val_loss: 0.0818 - lr: 1.0000e-05 - 453ms/epoch - 5ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7462 - val_loss: 0.0819 - lr: 1.0000e-05 - 453ms/epoch - 5ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7458 - val_loss: 0.0821 - lr: 1.0000e-05 - 453ms/epoch - 5ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.04545
90/90 - 1s - loss: 0.7454 - val_loss: 0.0822 - lr: 1.0000e-05 - 557ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.04545
90/90 - 1s - loss: 0.7450 - val_loss: 0.0823 - lr: 1.0000e-05 - 576ms/epoch - 6ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7446 - val_loss: 0.0825 - lr: 1.0000e-05 - 468ms/epoch - 5ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7442 - val_loss: 0.0826 - lr: 1.0000e-05 - 449ms/epoch - 5ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7438 - val_loss: 0.0827 - lr: 1.0000e-05 - 438ms/epoch - 5ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7434 - val_loss: 0.0829 - lr: 1.0000e-05 - 463ms/epoch - 5ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7430 - val_loss: 0.0830 - lr: 1.0000e-05 - 450ms/epoch - 5ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7426 - val_loss: 0.0832 - lr: 1.0000e-05 - 470ms/epoch - 5ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.04545
90/90 - 1s - loss: 0.7422 - val_loss: 0.0833 - lr: 1.0000e-05 - 541ms/epoch - 6ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7418 - val_loss: 0.0835 - lr: 1.0000e-05 - 433ms/epoch - 5ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7414 - val_loss: 0.0836 - lr: 1.0000e-05 - 480ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7410 - val_loss: 0.0838 - lr: 1.0000e-05 - 433ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7406 - val_loss: 0.0840 - lr: 1.0000e-05 - 452ms/epoch - 5ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.04545
90/90 - 1s - loss: 0.7402 - val_loss: 0.0841 - lr: 1.0000e-05 - 561ms/epoch - 6ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7397 - val_loss: 0.0843 - lr: 1.0000e-05 - 469ms/epoch - 5ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7393 - val_loss: 0.0845 - lr: 1.0000e-05 - 440ms/epoch - 5ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.04545
90/90 - 1s - loss: 0.7389 - val_loss: 0.0846 - lr: 1.0000e-05 - 556ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7385 - val_loss: 0.0848 - lr: 1.0000e-05 - 437ms/epoch - 5ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.04545
90/90 - 1s - loss: 0.7381 - val_loss: 0.0850 - lr: 1.0000e-05 - 545ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7377 - val_loss: 0.0852 - lr: 1.0000e-05 - 433ms/epoch - 5ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7373 - val_loss: 0.0853 - lr: 1.0000e-05 - 464ms/epoch - 5ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7369 - val_loss: 0.0855 - lr: 1.0000e-05 - 431ms/epoch - 5ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.04545
90/90 - 1s - loss: 0.7365 - val_loss: 0.0857 - lr: 1.0000e-05 - 520ms/epoch - 6ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7361 - val_loss: 0.0859 - lr: 1.0000e-05 - 447ms/epoch - 5ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7357 - val_loss: 0.0861 - lr: 1.0000e-05 - 446ms/epoch - 5ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.04545
90/90 - 0s - loss: 0.7353 - val_loss: 0.0863 - lr: 1.0000e-05 - 470ms/epoch - 5ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.04545
90/90 - 1s - loss: 0.7349 - val_loss: 0.0865 - lr: 1.0000e-05 - 510ms/epoch - 6ms/step
Epoch 00051: early stopping
SMA Prediction vs Close: 54.1% Accuracy Prediction vs Prediction: 51.87% Accuracy MSE: 25.081351909450348 RMSE: 5.008128583557969 MAPE: 3.9377037384058786 EMA Prediction vs Close: 54.1% Accuracy Prediction vs Prediction: 47.01% Accuracy MSE: 37.86506779592453 RMSE: 6.153459823215273 MAPE: 4.830217084369749 WMA Prediction vs Close: 53.36% Accuracy Prediction vs Prediction: 47.76% Accuracy MSE: 53.37812690087662 RMSE: 7.306033595657539 MAPE: 5.95316326588041 DEMA Prediction vs Close: 52.61% Accuracy Prediction vs Prediction: 44.78% Accuracy MSE: 133.19678136227444 RMSE: 11.541090995320783 MAPE: 10.29859546777107 KAMA Prediction vs Close: 54.48% Accuracy Prediction vs Prediction: 50.37% Accuracy MSE: 20.693935177088164 RMSE: 4.549058713304123 MAPE: 3.6577262429810227 MIDPOINT Prediction vs Close: 48.13% Accuracy Prediction vs Prediction: 47.01% Accuracy MSE: 18.24421544500263 RMSE: 4.271324788049093 MAPE: 3.3887721441386436 T3 Prediction vs Close: 53.73% Accuracy Prediction vs Prediction: 46.64% Accuracy MSE: 74.9216743993189 RMSE: 8.655730725901707 MAPE: 7.03412901576443 TEMA Prediction vs Close: 51.87% Accuracy Prediction vs Prediction: 48.88% Accuracy MSE: 33.63623508699449 RMSE: 5.799675429452453 MAPE: 5.291282498785388 Runtime: mins: 11.956794810899995
Architecture Used¶
In [108]:
from google.colab import files
import cv2
uploaded = files.upload()
In [109]:
img = cv2.imread('Experiment4.png')
plt.figure(figsize=(20,10))
plt.axis("off")
plt.title('LSTM Architecture '+imgfile,fontsize=18)
plt.imshow(img)
Out[109]:
<matplotlib.image.AxesImage at 0x7f7663294ad0>
Model Plots¶
In [103]:
with open('simulation4_data.json') as json_file:
simulation4 = json.load(json_file)
fileimg = 'Experiment4'
In [104]:
for i in range(len(list(simulation4.keys()))):
SIM = list(simulation4.keys())[i]
plot_train(simulation4,SIM)
plot_test(simulation4,SIM)
----- Train RMSE for SMA ----- 2.1639669204351253 ----- Train_MSE_LSTM for SMA ----- 4.68275283273748 ----- Train MAE LSTM for SMA ----- 2.1075676549779305
----- Test RMSE for SMA----- 5.008128583557969 ----- Test_MSE_LSTM for SMA----- 25.081351909450348 ----- Test_MAE_LSTM for SMA----- 3.9377037384058786
----- Train RMSE for EMA ----- 5.204058044811815 ----- Train_MSE_LSTM for EMA ----- 27.082220133770573 ----- Train MAE LSTM for EMA ----- 5.18856418958985
----- Test RMSE for EMA----- 6.153459823215273 ----- Test_MSE_LSTM for EMA----- 37.86506779592453 ----- Test_MAE_LSTM for EMA----- 4.830217084369749
----- Train RMSE for WMA ----- 2.3492817584355308 ----- Train_MSE_LSTM for WMA ----- 5.519124780517939 ----- Train MAE LSTM for WMA ----- 2.050265444387304
----- Test RMSE for WMA----- 7.306033595657539 ----- Test_MSE_LSTM for WMA----- 53.37812690087662 ----- Test_MAE_LSTM for WMA----- 5.95316326588041
----- Train RMSE for DEMA ----- 5.563977433104181 ----- Train_MSE_LSTM for DEMA ----- 30.957844876092594 ----- Train MAE LSTM for DEMA ----- 5.534693951653962
----- Test RMSE for DEMA----- 11.541090995320783 ----- Test_MSE_LSTM for DEMA----- 133.19678136227444 ----- Test_MAE_LSTM for DEMA----- 10.29859546777107
----- Train RMSE for KAMA ----- 1.9630444354122671 ----- Train_MSE_LSTM for KAMA ----- 3.8535434554030665 ----- Train MAE LSTM for KAMA ----- 1.943726107625678
----- Test RMSE for KAMA----- 4.549058713304123 ----- Test_MSE_LSTM for KAMA----- 20.693935177088164 ----- Test_MAE_LSTM for KAMA----- 3.6577262429810227
----- Train RMSE for MIDPOINT ----- 4.428587416955797 ----- Train_MSE_LSTM for MIDPOINT ----- 19.612386509619217 ----- Train MAE LSTM for MIDPOINT ----- 4.4129323463628785
----- Test RMSE for MIDPOINT----- 4.271324788049093 ----- Test_MSE_LSTM for MIDPOINT----- 18.24421544500263 ----- Test_MAE_LSTM for MIDPOINT----- 3.3887721441386436
----- Train RMSE for T3 ----- 3.3089997289158237 ----- Train_MSE_LSTM for T3 ----- 10.949479205964995 ----- Train MAE LSTM for T3 ----- 3.209559452415693
----- Test RMSE for T3----- 8.655730725901707 ----- Test_MSE_LSTM for T3----- 74.9216743993189 ----- Test_MAE_LSTM for T3----- 7.03412901576443
----- Train RMSE for TEMA ----- 1.0554541673050564 ----- Train_MSE_LSTM for TEMA ----- 1.1139834992816098 ----- Train MAE LSTM for TEMA ----- 0.602800428277195
----- Test RMSE for TEMA----- 5.799675429452453 ----- Test_MSE_LSTM for TEMA----- 33.63623508699449 ----- Test_MAE_LSTM for TEMA----- 5.291282498785388
Arima w Exogenous Variable Multistep MutiVariate LSTM Hybrid Model Experiment 5¶
In [114]:
def get_arima_exog(dataframe,original_data, train_len, test_len):
# prepare train and test data for exogenous vr
X_value = pd.DataFrame(low_vol.iloc[:, :])
y_value = pd.DataFrame(low_vol.iloc[:, 3])
X_scaler = MinMaxScaler(feature_range=(-1, 1))
y_scaler = MinMaxScaler(feature_range=(-1, 1))
X_scaler.fit(X_value)
y_scaler.fit(y_value)
X_scale_dataset = X_scaler.fit_transform(X_value)
y_scale_dataset = y_scaler.fit_transform(y_value)
# Get data and check shape
# X, y, yc = get_X_y(X_scale_dataset, y_scale_dataset)#X will be of shape 224 X 3 X 21 (each 3 X 21 array will be 3 days' worth of data). yc will have the corresponding closing price value
# pdb.set_trace()
X_train, X_test, = split_train_test(X_scale_dataset)
y_train, y_test, = split_train_test(y_scale_dataset)
yc_train,yc_test = split_train_test(low_vol_data)
yc = yc_test.values.tolist()
y_train_list = y_train.flatten().tolist()
y_test_list = y_test.flatten().tolist()
# yc_train, yc_test, = split_train_test(original_data)
index_train, index_test, = predict_index(dataset_final, X_train, n_steps_in, n_steps_out)
# Initialize model
model = auto_arima(y_train_list,exogenous = X_train,trace=True, error_action='ignore', start_p=1,start_q=1,max_p=3,max_q=3,d=3,
suppress_warnings=True,stepwise=True,seasonal=True)
# Determine model parameters
print(model.summary())
model.fit(y_train_list,maxiter=200)
order = model.get_params()['order']
print('ARIMA order:', order, '\n')
# Genereate predictions
prediction = []
for i in range(len(y_test_list)):
model = pmdarima.ARIMA(order=order)
model.fit(y_train_list)
# print('working on', i+1, 'of', len(y_test), '-- ' + str(int(100 * (i + 1) / len(y_test))) + '% complete')
prediction.append(model.predict()[0])
y_train_list.append(y_test_list[i])
predictionte = y_scaler.inverse_transform(np.array(prediction).reshape(-1,1))
y_test_ = y_scaler.inverse_transform(np.array(y_test_list).reshape(-1,1))
# Generate error data
mse = mean_squared_error(yc_test, predictionte)
rmse = mse ** 0.5
mae = mean_absolute_error(y_test_ , predictionte )
return yc,predictionte.flatten().tolist(), mse, rmse, mae
In [115]:
def get_lstm(data,original_data, train_len, test_len,img_file,ma ,lstm_len=3):
# prepare train and test data
X_value = pd.DataFrame(data.iloc[:, :])
y_value = pd.DataFrame(data.iloc[:, 3])
X_scaler = MinMaxScaler(feature_range=(-1, 1))
y_scaler = MinMaxScaler(feature_range=(-1, 1))
X_scaler.fit(X_value)
y_scaler.fit(y_value)
# Get data and check shape
X, y, yc = get_X_y(X_scale_dataset, y_scale_dataset)#X will be of shape 224 X 3 X 21 (each 3 X 21 array will be 3 days' worth of data). yc will have the corresponding closing price value
# pdb.set_trace()
X_train, X_test, = split_train_test(X)
y_train, y_test, = split_train_test(y)
# yc_train, yc_test, = split_train_test(original_data)
index_train, index_test, = predict_index(dataset_final, X_train, n_steps_in, n_steps_out)
det = 20
input_dim = X_train.shape[1]#3
feature_size = X_train.shape[2]#24
output_dim = y_train.shape[1]#1
# Option 1
# Set up & fit LSTM RNN
model = Sequential()
model.add(LSTM(256, activation='relu', kernel_initializer='he_normal', input_shape=(input_dim, feature_size)))
model.add(Dense(units=64,activation='relu'))
model.add(Dropout(0.5))
model.add(Dense(units=output_dim))
model.compile(optimizer=Adam(learning_rate = 0.001), loss='mse')
## Common code
callbacks = [
EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
ModelCheckpoint('LSTM5.h5', verbose=1, save_best_only=True, save_weights_only=True)]
fname1 = img_file+'.png'
tensorflow.keras.utils.plot_model(
model, to_file=fname1, show_shapes=True, show_dtype=False,
show_layer_names=True, expand_nested=False, dpi=96,
layer_range=None, show_layer_activations=False
)
history = model.fit(X_train, y_train, epochs=500, batch_size=int( optimized_period[ma]), verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# plot loss
fname2 = img_file+'-'+ma
plt.title(img_file+'-'+ma+' Loss')
plt.xlabel("Epochs")
plt.ylabel("Loss")
pyplot.plot(history.history['loss'], label='train')
pyplot.plot(history.history['val_loss'], label='validation')
pyplot.legend()
pyplot.savefig(fname2+'.png',dpi='figure')
pyplot.show()
# # option 2
# model = Sequential()
# model.add(Bidirectional(LSTM(units= 128), input_shape=(input_dim, feature_size)))
# model.add(Dense(64))
# model.add(Dense(units=output_dim))
# model.compile(optimizer=Adam(lr = 0.001), loss='mean_squared_error', metrics=['accuracy'])
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# Option 3
# define custom activation
#
# class Double_Tanh(Activation):
# def __init__(self, activation, **kwargs):
# super(Double_Tanh, self).__init__(activation, **kwargs)
# self.__name__ = 'double_tanh'
# def double_tanh(x):
# return (K.tanh(x) * 2)
# get_custom_objects().update({'double_tanh':Double_Tanh(double_tanh)})
# # Model Generation
# model = Sequential()
# #check https://machinelearningmastery.com/use-weight-regularization-lstm-networks-time-series-forecasting/
# model.add(LSTM(25, input_shape=(input_dim, feature_size), dropout=0.2, kernel_regularizer=l1_l2(0.00,0.00), bias_regularizer=l1_l2(0.00,0.00)))
# model.add(Dense(1))
# model.add(Activation(double_tanh))
# model.compile(loss='mean_squared_error', optimizer='adam', metrics=['mse', 'mae'])
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# Option 4
# Set up & fit LSTM RNN
# model = Sequential()
# model.add(LSTM(units=lstm_len, return_sequences=True, input_shape=(input_dim, feature_size)))
# model.add(LSTM(units=int(lstm_len/2)))
# model.add(Dense(1, activation='sigmoid'))
# model.compile(loss='mean_squared_error', optimizer='adam')
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM5.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# Generate predictions
predictiontr = model.predict(X_train, verbose=0)
predictiontr = y_scaler.inverse_transform(predictiontr).tolist()
outputtr = []
for i in range(len(predictiontr)):
outputtr.extend(predictiontr[i])
predictiontr = outputtr
# Generate error data
## replace with yc , xtest generated by new multistep method
mse_tr = mean_squared_error(y_train, predictiontr)
rmse_tr = mse_tr ** 0.5
# mape = mean_absolute_percentage_error(X_test, pd.Series(predictiontr))
mae_tr = mean_absolute_error(y_train, pd.Series(predictiontr))
# Original_tr = pd.Series(yc_train)
Original_tr = y_scaler.inverse_transform(y_train).flatten().tolist()
predictionte = model.predict(X_test, verbose=0)
predictionte = (y_scaler.inverse_transform(predictionte)-det).tolist()
outputte = []
for i in range(len(predictionte)):
outputte.extend(predictionte[i])
predictionte = outputte
# Generate error data
mse_te = mean_squared_error(y_test, predictionte)
rmse_te = mse_te ** 0.5
# mape = mean_absolute_percentage_error(X_test, pd.Series(predictionte))
mae_te = mean_absolute_error(y_test, pd.Series(predictionte))
# Original_te = pd.Series(yc_test)
Original_te = y_scaler.inverse_transform(y_test).flatten().tolist()
return Original_tr, predictiontr, mse_tr, rmse_tr,mae_tr,Original_te,predictionte, mse_te,rmse_te,mae_te
In [116]:
if __name__ == '__main__':
start_time = timeit.default_timer()
simulation5 = {}
imgfile = 'Experiment5'
for ma in optimized_period:
print(ma)
print(functions[ma])
print ( int( optimized_period[ma]))
# if ma == 'SMA':
low_vol = df.apply(lambda c: functions[ma](c, timeperiod = int( optimized_period[ma])))
low_vol = low_vol.fillna(0)
low_vol_data = df['close']
high_vol = pd.DataFrame()
df2 = df.copy()
for i in df2.columns:
if i in low_vol.columns:
high_vol[i] = df2[i].subtract(low_vol[i], fill_value=0)
high_vol_data = df['close']
## *****************************************************
# Generate ARIMA and LSTM predictions
print('\nWorking on ' + ma + ' predictions')
try:
print('parameters used : ', train_len, test_len)
low_vol_Original, low_vol_prediction, low_vol_mse, low_vol_rmse,low_vol_mae = get_arima_exog(low_vol,low_vol_data, train_len, test_len)
except:
print('ARIMA error, skipping to next MA type')
continue
Original_tr, predictiontr, mse_tr, rmse_tr,mae_tr,high_vol_Original, high_vol_prediction, high_vol_mse, high_vol_rmse,high_vol_mae, = get_lstm(high_vol,high_vol_data, train_len, test_len,imgfile,ma)
final_prediction_tr = df['close'].head(train_len).values + pd.Series(predictiontr) # ignoring first 3 steps
mse_ftr = mean_squared_error(df['close'].head(train_len).values,final_prediction_tr.values)
rmse_ftr = mse_ftr ** 0.5
mape_ftr = mean_absolute_percentage_error(df['close'].head(train_len).reset_index(drop=True), final_prediction_tr)
mae_ftr = mean_absolute_error(df['close'].head(train_len).reset_index(drop=True), final_prediction_tr)
final_prediction = pd.Series(low_vol_prediction[3:]) + pd.Series(high_vol_prediction)
mse = mean_squared_error(df['close'].tail(test_len).values,final_prediction.values)
rmse = mse ** 0.5
mape = mean_absolute_percentage_error(df['close'].tail(test_len).reset_index(drop=True), final_prediction)
mae = mean_absolute_error(df['close'].tail(test_len).reset_index(drop=True), final_prediction)
# Generate prediction accuracy
actual = df['close'].tail(test_len).values
result_1 = []
result_2 = []
for i in range(1, len(final_prediction)):
# Compare prediction to previous close price
if final_prediction[i] > actual[i-1] and actual[i] > actual[i-1]:
result_1.append(1)
elif final_prediction[i] < actual[i-1] and actual[i] < actual[i-1]:
result_1.append(1)
else:
result_1.append(0)
# Compare prediction to previous prediction
if final_prediction[i] > final_prediction[i-1] and actual[i] > actual[i-1]:
result_2.append(1)
elif final_prediction[i] < final_prediction[i-1] and actual[i] < actual[i-1]:
result_2.append(1)
else:
result_2.append(0)
accuracy_1 = np.mean(result_1)
accuracy_2 = np.mean(result_2)
simulation5[ma] = {'low_vol': {'original':list(low_vol_Original), 'prediction': list(low_vol_prediction) , 'mse': low_vol_mse,
'rmse': low_vol_rmse, 'mae' : low_vol_mae},
'high_vol': {'original':list(high_vol_Original),'prediction': list(high_vol_prediction), 'mse': high_vol_mse,
'rmse': high_vol_rmse, 'mae' : high_vol_mae},
'final_tr': {'original':df['close'].head(train_len).tolist(),'prediction': final_prediction_tr.values.tolist(), 'mse': mse_ftr,
'rmse': rmse_ftr, 'mae' : mae_ftr},
'final': {'original': df['close'].tail(test_len).tolist(), 'prediction': final_prediction.values.tolist(), 'mse': mse,
'rmse': rmse, 'mae': mae },
'accuracy': {'prediction vs close': accuracy_1, 'prediction vs prediction': accuracy_2}}
# save simulation data here as checkpoint
with open('simulation5_data.json', 'w') as fp:
json.dump(simulation5, fp)
for ma in simulation5.keys():
print('\n' + ma)
print('Prediction vs Close:\t\t' + str(round(100*simulation5[ma]['accuracy']['prediction vs close'], 2))
+ '% Accuracy')
print('Prediction vs Prediction:\t' + str(round(100*simulation5[ma]['accuracy']['prediction vs prediction'], 2))
+ '% Accuracy')
print('MSE:\t', simulation5[ma]['final']['mse'],
'\nRMSE:\t', simulation5[ma]['final']['rmse'],
'\nMAPE:\t', simulation5[ma]['final']['mae'])#,
# '\nMAPE:\t', simulation[ma]['final']['mape'])
# else:
# break
elapsed = timeit.default_timer() - start_time
print('Runtime: mins:',elapsed/60)
SMA
SMA([input_arrays], [timeperiod=30])
Simple Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
17
Working on SMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.786, Time=3.27 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.592, Time=4.65 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-15578.394, Time=8.47 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.592, Time=6.72 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16966.361, Time=9.33 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-16121.635, Time=9.70 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17214.069, Time=13.01 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-14562.592, Time=9.54 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-14572.319, Time=9.31 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-14403.474, Time=41.51 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 115.521 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8634.035
Date: Sun, 12 Dec 2021 AIC -17214.069
Time: 15:07:42 BIC -17087.416
Sample: 0 HQIC -17165.429
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -4.257e-09 9.55e-06 -0.000 1.000 -1.87e-05 1.87e-05
x2 -4.256e-09 9.56e-06 -0.000 1.000 -1.87e-05 1.87e-05
x3 -4.313e-09 9.62e-06 -0.000 1.000 -1.89e-05 1.88e-05
x4 1.0000 9.61e-06 1.04e+05 0.000 1.000 1.000
x5 -3.891e-09 9.14e-06 -0.000 1.000 -1.79e-05 1.79e-05
x6 -1.122e-08 1.03e-05 -0.001 0.999 -2.03e-05 2.03e-05
x7 -4.223e-09 9.54e-06 -0.000 1.000 -1.87e-05 1.87e-05
x8 -4.234e-09 9.55e-06 -0.000 1.000 -1.87e-05 1.87e-05
x9 -1.626e-10 6.54e-07 -0.000 1.000 -1.28e-06 1.28e-06
x10 -6.831e-10 2.91e-06 -0.000 1.000 -5.7e-06 5.7e-06
x11 -4.115e-09 9.41e-06 -0.000 1.000 -1.84e-05 1.84e-05
x12 -4.303e-09 9.62e-06 -0.000 1.000 -1.89e-05 1.88e-05
x13 -4.288e-09 9.6e-06 -0.000 1.000 -1.88e-05 1.88e-05
x14 -3.749e-08 2.81e-05 -0.001 0.999 -5.51e-05 5.5e-05
x15 -5.032e-09 1.04e-05 -0.000 1.000 -2.04e-05 2.03e-05
x16 -3.685e-09 9e-06 -0.000 1.000 -1.76e-05 1.76e-05
x17 -3.286e-09 8.45e-06 -0.000 1.000 -1.66e-05 1.66e-05
x18 -1.22e-08 1.59e-05 -0.001 0.999 -3.11e-05 3.11e-05
x19 -5.685e-09 1.1e-05 -0.001 1.000 -2.16e-05 2.16e-05
x20 -1.42e-08 1.69e-05 -0.001 0.999 -3.32e-05 3.32e-05
x21 -5.194e-08 3.31e-05 -0.002 0.999 -6.49e-05 6.48e-05
x22 -2.548e-08 2.31e-05 -0.001 0.999 -4.53e-05 4.52e-05
x23 -3.534e-08 2.73e-05 -0.001 0.999 -5.35e-05 5.34e-05
x24 -1.566e-08 1.8e-05 -0.001 0.999 -3.53e-05 3.53e-05
ma.L1 -1.3899 4.98e-09 -2.79e+08 0.000 -1.390 -1.390
ma.L2 0.4032 4.98e-09 8.09e+07 0.000 0.403 0.403
sigma2 7.635e-11 6.92e-11 1.103 0.270 -5.93e-11 2.12e-10
===================================================================================
Ljung-Box (L1) (Q): 68.48 Jarque-Bera (JB): 5579791.06
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 10.12
Prob(H) (two-sided): 0.00 Kurtosis: 410.36
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.69e+24. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
WARNING:tensorflow:Layer lstm_49 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
WARNING:tensorflow:Layer lstm_49 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.56547, saving model to LSTM5.h5 48/48 - 2s - loss: 0.2090 - val_loss: 0.5655 - lr: 0.0010 - 2s/epoch - 43ms/step Epoch 2/500 Epoch 00002: val_loss improved from 0.56547 to 0.03066, saving model to LSTM5.h5 48/48 - 0s - loss: 0.1069 - val_loss: 0.0307 - lr: 0.0010 - 405ms/epoch - 8ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.03066 48/48 - 0s - loss: 0.2029 - val_loss: 0.6358 - lr: 0.0010 - 454ms/epoch - 9ms/step Epoch 4/500 Epoch 00004: val_loss improved from 0.03066 to 0.02836, saving model to LSTM5.h5 48/48 - 0s - loss: 0.0706 - val_loss: 0.0284 - lr: 0.0010 - 400ms/epoch - 8ms/step Epoch 5/500 Epoch 00005: val_loss did not improve from 0.02836 48/48 - 0s - loss: 0.0601 - val_loss: 0.1206 - lr: 0.0010 - 448ms/epoch - 9ms/step Epoch 6/500 Epoch 00006: val_loss improved from 0.02836 to 0.01619, saving model to LSTM5.h5 48/48 - 0s - loss: 0.0450 - val_loss: 0.0162 - lr: 0.0010 - 449ms/epoch - 9ms/step Epoch 7/500 Epoch 00007: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0492 - val_loss: 0.1798 - lr: 0.0010 - 434ms/epoch - 9ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0346 - val_loss: 0.0218 - lr: 0.0010 - 432ms/epoch - 9ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0408 - val_loss: 0.2343 - lr: 0.0010 - 418ms/epoch - 9ms/step Epoch 10/500 Epoch 00010: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0355 - val_loss: 0.0463 - lr: 0.0010 - 464ms/epoch - 10ms/step Epoch 11/500 Epoch 00011: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00011: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0305 - val_loss: 0.1022 - lr: 0.0010 - 396ms/epoch - 8ms/step Epoch 12/500 Epoch 00012: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0334 - val_loss: 0.0843 - lr: 1.0000e-04 - 391ms/epoch - 8ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0300 - val_loss: 0.0847 - lr: 1.0000e-04 - 431ms/epoch - 9ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0297 - val_loss: 0.0868 - lr: 1.0000e-04 - 378ms/epoch - 8ms/step Epoch 15/500 Epoch 00015: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0281 - val_loss: 0.0946 - lr: 1.0000e-04 - 454ms/epoch - 9ms/step Epoch 16/500 Epoch 00016: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00016: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0260 - val_loss: 0.0970 - lr: 1.0000e-04 - 467ms/epoch - 10ms/step Epoch 17/500 Epoch 00017: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0265 - val_loss: 0.0971 - lr: 1.0000e-05 - 410ms/epoch - 9ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0264 - val_loss: 0.0966 - lr: 1.0000e-05 - 426ms/epoch - 9ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0267 - val_loss: 0.0955 - lr: 1.0000e-05 - 427ms/epoch - 9ms/step Epoch 20/500 Epoch 00020: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0287 - val_loss: 0.0951 - lr: 1.0000e-05 - 424ms/epoch - 9ms/step Epoch 21/500 Epoch 00021: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00021: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0279 - val_loss: 0.0953 - lr: 1.0000e-05 - 411ms/epoch - 9ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0270 - val_loss: 0.0948 - lr: 1.0000e-05 - 418ms/epoch - 9ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0254 - val_loss: 0.0947 - lr: 1.0000e-05 - 460ms/epoch - 10ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0258 - val_loss: 0.0947 - lr: 1.0000e-05 - 385ms/epoch - 8ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0253 - val_loss: 0.0947 - lr: 1.0000e-05 - 408ms/epoch - 9ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0270 - val_loss: 0.0947 - lr: 1.0000e-05 - 423ms/epoch - 9ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0281 - val_loss: 0.0948 - lr: 1.0000e-05 - 393ms/epoch - 8ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0256 - val_loss: 0.0942 - lr: 1.0000e-05 - 371ms/epoch - 8ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0281 - val_loss: 0.0945 - lr: 1.0000e-05 - 438ms/epoch - 9ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0253 - val_loss: 0.0948 - lr: 1.0000e-05 - 437ms/epoch - 9ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0268 - val_loss: 0.0939 - lr: 1.0000e-05 - 407ms/epoch - 8ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0269 - val_loss: 0.0939 - lr: 1.0000e-05 - 423ms/epoch - 9ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0308 - val_loss: 0.0934 - lr: 1.0000e-05 - 420ms/epoch - 9ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0248 - val_loss: 0.0917 - lr: 1.0000e-05 - 399ms/epoch - 8ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0268 - val_loss: 0.0921 - lr: 1.0000e-05 - 411ms/epoch - 9ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0255 - val_loss: 0.0919 - lr: 1.0000e-05 - 380ms/epoch - 8ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0273 - val_loss: 0.0924 - lr: 1.0000e-05 - 421ms/epoch - 9ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0256 - val_loss: 0.0929 - lr: 1.0000e-05 - 395ms/epoch - 8ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0292 - val_loss: 0.0931 - lr: 1.0000e-05 - 407ms/epoch - 8ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0269 - val_loss: 0.0930 - lr: 1.0000e-05 - 368ms/epoch - 8ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0262 - val_loss: 0.0935 - lr: 1.0000e-05 - 438ms/epoch - 9ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0260 - val_loss: 0.0936 - lr: 1.0000e-05 - 486ms/epoch - 10ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0271 - val_loss: 0.0910 - lr: 1.0000e-05 - 414ms/epoch - 9ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0252 - val_loss: 0.0918 - lr: 1.0000e-05 - 384ms/epoch - 8ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0250 - val_loss: 0.0919 - lr: 1.0000e-05 - 398ms/epoch - 8ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0259 - val_loss: 0.0921 - lr: 1.0000e-05 - 420ms/epoch - 9ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0236 - val_loss: 0.0919 - lr: 1.0000e-05 - 408ms/epoch - 9ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0249 - val_loss: 0.0914 - lr: 1.0000e-05 - 399ms/epoch - 8ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0268 - val_loss: 0.0923 - lr: 1.0000e-05 - 421ms/epoch - 9ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0233 - val_loss: 0.0933 - lr: 1.0000e-05 - 393ms/epoch - 8ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0261 - val_loss: 0.0927 - lr: 1.0000e-05 - 393ms/epoch - 8ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0271 - val_loss: 0.0922 - lr: 1.0000e-05 - 402ms/epoch - 8ms/step Epoch 53/500 Epoch 00053: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0252 - val_loss: 0.0903 - lr: 1.0000e-05 - 402ms/epoch - 8ms/step Epoch 54/500 Epoch 00054: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0269 - val_loss: 0.0911 - lr: 1.0000e-05 - 403ms/epoch - 8ms/step Epoch 55/500 Epoch 00055: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0258 - val_loss: 0.0920 - lr: 1.0000e-05 - 377ms/epoch - 8ms/step Epoch 56/500 Epoch 00056: val_loss did not improve from 0.01619 48/48 - 0s - loss: 0.0264 - val_loss: 0.0917 - lr: 1.0000e-05 - 421ms/epoch - 9ms/step Epoch 00056: early stopping
SMA
Prediction vs Close: 46.27% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 22.905457367129987
RMSE: 4.785964622427749
MAPE: 4.003866138329542
EMA
EMA([input_arrays], [timeperiod=30])
Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
51
Working on EMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.778, Time=3.17 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.589, Time=4.74 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-14606.447, Time=6.20 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.589, Time=6.93 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-15343.613, Time=10.18 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-15047.583, Time=13.32 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-16858.964, Time=11.76 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=-17024.022, Time=5.75 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=-16998.618, Time=3.58 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=-17081.451, Time=6.55 sec
ARIMA(3,3,2)(0,0,0)[0] : AIC=inf, Time=16.66 sec
/usr/local/lib/python3.7/dist-packages/statsmodels/tsa/statespace/sarimax.py:1906: RuntimeWarning: divide by zero encountered in reciprocal return np.roots(self.polynomial_reduced_ma)**-1
ARIMA(2,3,2)(0,0,0)[0] : AIC=-16997.990, Time=3.47 sec
ARIMA(3,3,1)(0,0,0)[0] intercept : AIC=-16992.667, Time=4.34 sec
Best model: ARIMA(3,3,1)(0,0,0)[0]
Total fit time: 96.671 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 1) Log Likelihood 8569.726
Date: Sun, 12 Dec 2021 AIC -17081.451
Time: 15:13:34 BIC -16945.417
Sample: 0 HQIC -17029.208
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -2.316e-10 9.89e-05 -2.34e-06 1.000 -0.000 0.000
x2 -2.309e-10 9.88e-05 -2.34e-06 1.000 -0.000 0.000
x3 -2.325e-10 9.91e-05 -2.35e-06 1.000 -0.000 0.000
x4 1.0000 9.9e-05 1.01e+04 0.000 1.000 1.000
x5 -2.108e-10 9.43e-05 -2.24e-06 1.000 -0.000 0.000
x6 -7.997e-10 0.000 -4.63e-06 1.000 -0.000 0.000
x7 -2.295e-10 9.85e-05 -2.33e-06 1.000 -0.000 0.000
x8 -2.244e-10 9.74e-05 -2.3e-06 1.000 -0.000 0.000
x9 -1.166e-11 1.98e-05 -5.9e-07 1.000 -3.87e-05 3.87e-05
x10 -4.454e-11 4.19e-05 -1.06e-06 1.000 -8.22e-05 8.22e-05
x11 -2.219e-10 9.68e-05 -2.29e-06 1.000 -0.000 0.000
x12 -2.264e-10 9.8e-05 -2.31e-06 1.000 -0.000 0.000
x13 -2.315e-10 9.89e-05 -2.34e-06 1.000 -0.000 0.000
x14 -1.767e-09 0.000 -6.47e-06 1.000 -0.001 0.001
x15 -2.096e-10 9.38e-05 -2.23e-06 1.000 -0.000 0.000
x16 -5.257e-10 0.000 -3.5e-06 1.000 -0.000 0.000
x17 -2.143e-10 9.53e-05 -2.25e-06 1.000 -0.000 0.000
x18 -3.776e-11 3.61e-05 -1.05e-06 1.000 -7.08e-05 7.08e-05
x19 -2.52e-10 0.000 -2.41e-06 1.000 -0.000 0.000
x20 -2.417e-10 9.51e-05 -2.54e-06 1.000 -0.000 0.000
x21 -3.16e-09 0.000 -8.64e-06 1.000 -0.001 0.001
x22 -2.955e-09 0.000 -8.32e-06 1.000 -0.001 0.001
x23 -1.664e-09 0.000 -6.29e-06 1.000 -0.001 0.001
x24 -1.568e-09 0.000 -6.07e-06 1.000 -0.001 0.001
ar.L1 -0.4923 1.2e-09 -4.09e+08 0.000 -0.492 -0.492
ar.L2 -0.1923 7e-10 -2.75e+08 0.000 -0.192 -0.192
ar.L3 -0.0461 3.32e-10 -1.39e+08 0.000 -0.046 -0.046
ma.L1 -0.7077 2.73e-09 -2.59e+08 0.000 -0.708 -0.708
sigma2 8.99e-11 6.96e-11 1.291 0.197 -4.66e-11 2.26e-10
===================================================================================
Ljung-Box (L1) (Q): 55.51 Jarque-Bera (JB): 4268313.90
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 5.44
Prob(H) (two-sided): 0.00 Kurtosis: 359.56
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.36e+28. Standard errors may be unstable.
ARIMA order: (3, 3, 1)
WARNING:tensorflow:Layer lstm_50 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
WARNING:tensorflow:Layer lstm_50 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.21948, saving model to LSTM5.h5 16/16 - 2s - loss: 0.6907 - val_loss: 0.2195 - lr: 0.0010 - 2s/epoch - 130ms/step Epoch 2/500 Epoch 00002: val_loss improved from 0.21948 to 0.03167, saving model to LSTM5.h5 16/16 - 0s - loss: 0.2540 - val_loss: 0.0317 - lr: 0.0010 - 162ms/epoch - 10ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0860 - val_loss: 0.1915 - lr: 0.0010 - 161ms/epoch - 10ms/step Epoch 4/500 Epoch 00004: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0661 - val_loss: 0.1009 - lr: 0.0010 - 163ms/epoch - 10ms/step Epoch 5/500 Epoch 00005: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0508 - val_loss: 0.0429 - lr: 0.0010 - 153ms/epoch - 10ms/step Epoch 6/500 Epoch 00006: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0497 - val_loss: 0.1203 - lr: 0.0010 - 140ms/epoch - 9ms/step Epoch 7/500 Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00007: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0461 - val_loss: 0.1094 - lr: 0.0010 - 154ms/epoch - 10ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0424 - val_loss: 0.1108 - lr: 1.0000e-04 - 139ms/epoch - 9ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0430 - val_loss: 0.1112 - lr: 1.0000e-04 - 158ms/epoch - 10ms/step Epoch 10/500 Epoch 00010: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0397 - val_loss: 0.1123 - lr: 1.0000e-04 - 138ms/epoch - 9ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0406 - val_loss: 0.1096 - lr: 1.0000e-04 - 147ms/epoch - 9ms/step Epoch 12/500 Epoch 00012: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00012: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0431 - val_loss: 0.1044 - lr: 1.0000e-04 - 154ms/epoch - 10ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0408 - val_loss: 0.1044 - lr: 1.0000e-05 - 174ms/epoch - 11ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0399 - val_loss: 0.1044 - lr: 1.0000e-05 - 163ms/epoch - 10ms/step Epoch 15/500 Epoch 00015: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0426 - val_loss: 0.1045 - lr: 1.0000e-05 - 158ms/epoch - 10ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0406 - val_loss: 0.1052 - lr: 1.0000e-05 - 143ms/epoch - 9ms/step Epoch 17/500 Epoch 00017: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00017: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0433 - val_loss: 0.1052 - lr: 1.0000e-05 - 147ms/epoch - 9ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0407 - val_loss: 0.1047 - lr: 1.0000e-05 - 138ms/epoch - 9ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0416 - val_loss: 0.1043 - lr: 1.0000e-05 - 153ms/epoch - 10ms/step Epoch 20/500 Epoch 00020: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0422 - val_loss: 0.1041 - lr: 1.0000e-05 - 162ms/epoch - 10ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0394 - val_loss: 0.1038 - lr: 1.0000e-05 - 177ms/epoch - 11ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0391 - val_loss: 0.1034 - lr: 1.0000e-05 - 150ms/epoch - 9ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0392 - val_loss: 0.1031 - lr: 1.0000e-05 - 147ms/epoch - 9ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0398 - val_loss: 0.1035 - lr: 1.0000e-05 - 151ms/epoch - 9ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0357 - val_loss: 0.1032 - lr: 1.0000e-05 - 157ms/epoch - 10ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0371 - val_loss: 0.1032 - lr: 1.0000e-05 - 151ms/epoch - 9ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0396 - val_loss: 0.1033 - lr: 1.0000e-05 - 155ms/epoch - 10ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0390 - val_loss: 0.1028 - lr: 1.0000e-05 - 156ms/epoch - 10ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0421 - val_loss: 0.1028 - lr: 1.0000e-05 - 154ms/epoch - 10ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0409 - val_loss: 0.1021 - lr: 1.0000e-05 - 152ms/epoch - 9ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0422 - val_loss: 0.1018 - lr: 1.0000e-05 - 195ms/epoch - 12ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0390 - val_loss: 0.1017 - lr: 1.0000e-05 - 218ms/epoch - 14ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0424 - val_loss: 0.1012 - lr: 1.0000e-05 - 148ms/epoch - 9ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0418 - val_loss: 0.1011 - lr: 1.0000e-05 - 177ms/epoch - 11ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0394 - val_loss: 0.1007 - lr: 1.0000e-05 - 145ms/epoch - 9ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0395 - val_loss: 0.1006 - lr: 1.0000e-05 - 163ms/epoch - 10ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0377 - val_loss: 0.1003 - lr: 1.0000e-05 - 146ms/epoch - 9ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0432 - val_loss: 0.1001 - lr: 1.0000e-05 - 147ms/epoch - 9ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0366 - val_loss: 0.0997 - lr: 1.0000e-05 - 158ms/epoch - 10ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0355 - val_loss: 0.0997 - lr: 1.0000e-05 - 142ms/epoch - 9ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0424 - val_loss: 0.0994 - lr: 1.0000e-05 - 165ms/epoch - 10ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0404 - val_loss: 0.0990 - lr: 1.0000e-05 - 142ms/epoch - 9ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0370 - val_loss: 0.0990 - lr: 1.0000e-05 - 149ms/epoch - 9ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0396 - val_loss: 0.0997 - lr: 1.0000e-05 - 146ms/epoch - 9ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0402 - val_loss: 0.1002 - lr: 1.0000e-05 - 156ms/epoch - 10ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0368 - val_loss: 0.0999 - lr: 1.0000e-05 - 142ms/epoch - 9ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0410 - val_loss: 0.0996 - lr: 1.0000e-05 - 143ms/epoch - 9ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0392 - val_loss: 0.0985 - lr: 1.0000e-05 - 169ms/epoch - 11ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0401 - val_loss: 0.0980 - lr: 1.0000e-05 - 162ms/epoch - 10ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0356 - val_loss: 0.0976 - lr: 1.0000e-05 - 148ms/epoch - 9ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0376 - val_loss: 0.0983 - lr: 1.0000e-05 - 150ms/epoch - 9ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.03167 16/16 - 0s - loss: 0.0362 - val_loss: 0.0983 - lr: 1.0000e-05 - 147ms/epoch - 9ms/step Epoch 00052: early stopping
SMA
Prediction vs Close: 46.27% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 22.905457367129987
RMSE: 4.785964622427749
MAPE: 4.003866138329542
EMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 44.03% Accuracy
MSE: 24.329194491765858
RMSE: 4.932463328983385
MAPE: 4.099346553838579
WMA
WMA([input_arrays], [timeperiod=30])
Weighted Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
49
Working on WMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.780, Time=3.20 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.589, Time=4.50 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-16789.784, Time=12.14 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.589, Time=7.21 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16919.987, Time=9.59 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-14616.097, Time=12.21 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17225.955, Time=18.57 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-14562.589, Time=9.57 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-15582.364, Time=19.20 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-12043.670, Time=36.41 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 132.614 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8639.977
Date: Sun, 12 Dec 2021 AIC -17225.955
Time: 15:24:54 BIC -17099.302
Sample: 0 HQIC -17177.315
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -4.802e-09 4.51e-06 -0.001 0.999 -8.84e-06 8.83e-06
x2 -4.783e-09 4.5e-06 -0.001 0.999 -8.83e-06 8.82e-06
x3 -4.811e-09 4.51e-06 -0.001 0.999 -8.85e-06 8.84e-06
x4 1.0000 4.51e-06 2.22e+05 0.000 1.000 1.000
x5 -4.353e-09 4.3e-06 -0.001 0.999 -8.43e-06 8.42e-06
x6 -1.569e-08 7.54e-06 -0.002 0.998 -1.48e-05 1.48e-05
x7 -4.75e-09 4.49e-06 -0.001 0.999 -8.8e-06 8.79e-06
x8 -4.628e-09 4.43e-06 -0.001 0.999 -8.69e-06 8.69e-06
x9 -4.733e-10 1.16e-06 -0.000 1.000 -2.27e-06 2.27e-06
x10 -7.88e-10 1.8e-06 -0.000 1.000 -3.52e-06 3.52e-06
x11 -4.609e-09 4.42e-06 -0.001 0.999 -8.68e-06 8.67e-06
x12 -4.607e-09 4.42e-06 -0.001 0.999 -8.68e-06 8.67e-06
x13 -4.792e-09 4.51e-06 -0.001 0.999 -8.84e-06 8.83e-06
x14 -3.777e-08 1.24e-05 -0.003 0.998 -2.44e-05 2.44e-05
x15 -3.99e-09 4.12e-06 -0.001 0.999 -8.08e-06 8.07e-06
x16 -1.309e-08 7.41e-06 -0.002 0.999 -1.45e-05 1.45e-05
x17 -4.789e-09 4.51e-06 -0.001 0.999 -8.85e-06 8.84e-06
x18 -2.665e-10 9.77e-07 -0.000 1.000 -1.92e-06 1.92e-06
x19 -4.919e-09 4.56e-06 -0.001 0.999 -8.94e-06 8.93e-06
x20 -4e-10 9.58e-07 -0.000 1.000 -1.88e-06 1.88e-06
x21 -6.782e-08 1.67e-05 -0.004 0.997 -3.27e-05 3.26e-05
x22 -6.03e-08 1.58e-05 -0.004 0.997 -3.09e-05 3.08e-05
x23 -3.157e-08 1.14e-05 -0.003 0.998 -2.23e-05 2.23e-05
x24 -3.671e-08 1.23e-05 -0.003 0.998 -2.41e-05 2.41e-05
ma.L1 -1.3901 5.58e-10 -2.49e+09 0.000 -1.390 -1.390
ma.L2 0.4033 5.75e-10 7.02e+08 0.000 0.403 0.403
sigma2 7.525e-11 6.92e-11 1.088 0.277 -6.03e-11 2.11e-10
===================================================================================
Ljung-Box (L1) (Q): 69.18 Jarque-Bera (JB): 6366427.21
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 12.29
Prob(H) (two-sided): 0.00 Kurtosis: 437.97
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 2.29e+25. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
WARNING:tensorflow:Layer lstm_51 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
WARNING:tensorflow:Layer lstm_51 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.02747, saving model to LSTM5.h5 17/17 - 2s - loss: 0.2646 - val_loss: 0.0275 - lr: 0.0010 - 2s/epoch - 103ms/step Epoch 2/500 Epoch 00002: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.1004 - val_loss: 0.3699 - lr: 0.0010 - 139ms/epoch - 8ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0793 - val_loss: 0.1598 - lr: 0.0010 - 153ms/epoch - 9ms/step Epoch 4/500 Epoch 00004: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0976 - val_loss: 0.1705 - lr: 0.0010 - 168ms/epoch - 10ms/step Epoch 5/500 Epoch 00005: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0784 - val_loss: 0.0960 - lr: 0.0010 - 158ms/epoch - 9ms/step Epoch 6/500 Epoch 00006: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00006: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0657 - val_loss: 0.0759 - lr: 0.0010 - 186ms/epoch - 11ms/step Epoch 7/500 Epoch 00007: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0720 - val_loss: 0.0821 - lr: 1.0000e-04 - 161ms/epoch - 9ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0512 - val_loss: 0.0881 - lr: 1.0000e-04 - 150ms/epoch - 9ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0349 - val_loss: 0.0932 - lr: 1.0000e-04 - 154ms/epoch - 9ms/step Epoch 10/500 Epoch 00010: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0368 - val_loss: 0.0940 - lr: 1.0000e-04 - 163ms/epoch - 10ms/step Epoch 11/500 Epoch 00011: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00011: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0340 - val_loss: 0.0919 - lr: 1.0000e-04 - 149ms/epoch - 9ms/step Epoch 12/500 Epoch 00012: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0345 - val_loss: 0.0917 - lr: 1.0000e-05 - 149ms/epoch - 9ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0324 - val_loss: 0.0916 - lr: 1.0000e-05 - 152ms/epoch - 9ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0359 - val_loss: 0.0916 - lr: 1.0000e-05 - 156ms/epoch - 9ms/step Epoch 15/500 Epoch 00015: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0320 - val_loss: 0.0915 - lr: 1.0000e-05 - 149ms/epoch - 9ms/step Epoch 16/500 Epoch 00016: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00016: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0334 - val_loss: 0.0917 - lr: 1.0000e-05 - 172ms/epoch - 10ms/step Epoch 17/500 Epoch 00017: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0374 - val_loss: 0.0916 - lr: 1.0000e-05 - 165ms/epoch - 10ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0306 - val_loss: 0.0917 - lr: 1.0000e-05 - 166ms/epoch - 10ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0337 - val_loss: 0.0918 - lr: 1.0000e-05 - 156ms/epoch - 9ms/step Epoch 20/500 Epoch 00020: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0340 - val_loss: 0.0919 - lr: 1.0000e-05 - 207ms/epoch - 12ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0338 - val_loss: 0.0918 - lr: 1.0000e-05 - 152ms/epoch - 9ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0330 - val_loss: 0.0919 - lr: 1.0000e-05 - 169ms/epoch - 10ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0339 - val_loss: 0.0919 - lr: 1.0000e-05 - 160ms/epoch - 9ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0328 - val_loss: 0.0916 - lr: 1.0000e-05 - 162ms/epoch - 10ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0318 - val_loss: 0.0911 - lr: 1.0000e-05 - 150ms/epoch - 9ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0317 - val_loss: 0.0909 - lr: 1.0000e-05 - 160ms/epoch - 9ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0377 - val_loss: 0.0908 - lr: 1.0000e-05 - 147ms/epoch - 9ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0348 - val_loss: 0.0911 - lr: 1.0000e-05 - 160ms/epoch - 9ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0307 - val_loss: 0.0915 - lr: 1.0000e-05 - 162ms/epoch - 10ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0305 - val_loss: 0.0915 - lr: 1.0000e-05 - 154ms/epoch - 9ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0339 - val_loss: 0.0915 - lr: 1.0000e-05 - 143ms/epoch - 8ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0322 - val_loss: 0.0914 - lr: 1.0000e-05 - 167ms/epoch - 10ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0309 - val_loss: 0.0912 - lr: 1.0000e-05 - 150ms/epoch - 9ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0319 - val_loss: 0.0916 - lr: 1.0000e-05 - 153ms/epoch - 9ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0319 - val_loss: 0.0916 - lr: 1.0000e-05 - 177ms/epoch - 10ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0347 - val_loss: 0.0916 - lr: 1.0000e-05 - 167ms/epoch - 10ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0310 - val_loss: 0.0919 - lr: 1.0000e-05 - 162ms/epoch - 10ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0330 - val_loss: 0.0920 - lr: 1.0000e-05 - 163ms/epoch - 10ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0351 - val_loss: 0.0923 - lr: 1.0000e-05 - 150ms/epoch - 9ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0327 - val_loss: 0.0923 - lr: 1.0000e-05 - 165ms/epoch - 10ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0314 - val_loss: 0.0923 - lr: 1.0000e-05 - 163ms/epoch - 10ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0346 - val_loss: 0.0926 - lr: 1.0000e-05 - 153ms/epoch - 9ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0323 - val_loss: 0.0935 - lr: 1.0000e-05 - 156ms/epoch - 9ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0310 - val_loss: 0.0932 - lr: 1.0000e-05 - 176ms/epoch - 10ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0330 - val_loss: 0.0934 - lr: 1.0000e-05 - 156ms/epoch - 9ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0313 - val_loss: 0.0935 - lr: 1.0000e-05 - 155ms/epoch - 9ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0334 - val_loss: 0.0937 - lr: 1.0000e-05 - 161ms/epoch - 9ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0304 - val_loss: 0.0933 - lr: 1.0000e-05 - 148ms/epoch - 9ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0335 - val_loss: 0.0932 - lr: 1.0000e-05 - 167ms/epoch - 10ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0323 - val_loss: 0.0934 - lr: 1.0000e-05 - 151ms/epoch - 9ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.02747 17/17 - 0s - loss: 0.0295 - val_loss: 0.0933 - lr: 1.0000e-05 - 159ms/epoch - 9ms/step Epoch 00051: early stopping
SMA
Prediction vs Close: 46.27% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 22.905457367129987
RMSE: 4.785964622427749
MAPE: 4.003866138329542
EMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 44.03% Accuracy
MSE: 24.329194491765858
RMSE: 4.932463328983385
MAPE: 4.099346553838579
WMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 45.15% Accuracy
MSE: 26.731339639305343
RMSE: 5.170235936522176
MAPE: 4.142288801040536
DEMA
DEMA([input_arrays], [timeperiod=30])
Double Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
89
Working on DEMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.785, Time=3.45 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.588, Time=4.62 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-15575.689, Time=9.06 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.588, Time=7.23 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16714.796, Time=8.90 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-15610.140, Time=10.43 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17225.835, Time=22.86 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-14562.588, Time=8.58 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-16751.951, Time=20.65 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-11788.089, Time=30.91 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 126.719 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8639.917
Date: Sun, 12 Dec 2021 AIC -17225.835
Time: 15:31:29 BIC -17099.182
Sample: 0 HQIC -17177.195
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -5.894e-09 3.61e-06 -0.002 0.999 -7.09e-06 7.08e-06
x2 -5.93e-09 3.63e-06 -0.002 0.999 -7.11e-06 7.1e-06
x3 -5.905e-09 3.62e-06 -0.002 0.999 -7.1e-06 7.09e-06
x4 1.0000 3.62e-06 2.76e+05 0.000 1.000 1.000
x5 -5.457e-09 3.48e-06 -0.002 0.999 -6.83e-06 6.82e-06
x6 -3.019e-08 7.72e-06 -0.004 0.997 -1.52e-05 1.51e-05
x7 -5.87e-09 3.61e-06 -0.002 0.999 -7.08e-06 7.07e-06
x8 -5.809e-09 3.59e-06 -0.002 0.999 -7.05e-06 7.04e-06
x9 -9.293e-11 9.83e-08 -0.001 0.999 -1.93e-07 1.93e-07
x10 -2.793e-09 2.47e-06 -0.001 0.999 -4.84e-06 4.84e-06
x11 -6.095e-09 3.68e-06 -0.002 0.999 -7.21e-06 7.2e-06
x12 -5.478e-09 3.49e-06 -0.002 0.999 -6.85e-06 6.84e-06
x13 -5.91e-09 3.62e-06 -0.002 0.999 -7.1e-06 7.09e-06
x14 -4.085e-08 9.35e-06 -0.004 0.997 -1.84e-05 1.83e-05
x15 -5.93e-09 3.63e-06 -0.002 0.999 -7.12e-06 7.11e-06
x16 -1.618e-09 1.92e-06 -0.001 0.999 -3.76e-06 3.75e-06
x17 -5.076e-09 3.37e-06 -0.002 0.999 -6.6e-06 6.59e-06
x18 -1.377e-08 5.5e-06 -0.003 0.998 -1.08e-05 1.08e-05
x19 -6.135e-09 3.69e-06 -0.002 0.999 -7.23e-06 7.22e-06
x20 -1.018e-08 4.43e-06 -0.002 0.998 -8.68e-06 8.66e-06
x21 -6.911e-08 1.21e-05 -0.006 0.995 -2.39e-05 2.37e-05
x22 -5.656e-08 1.1e-05 -0.005 0.996 -2.16e-05 2.15e-05
x23 -5.355e-08 1.07e-05 -0.005 0.996 -2.1e-05 2.09e-05
x24 -3.636e-08 8.85e-06 -0.004 0.997 -1.74e-05 1.73e-05
ma.L1 -1.3899 4.86e-11 -2.86e+10 0.000 -1.390 -1.390
ma.L2 0.4032 4.6e-11 8.76e+09 0.000 0.403 0.403
sigma2 7.526e-11 6.92e-11 1.088 0.277 -6.03e-11 2.11e-10
===================================================================================
Ljung-Box (L1) (Q): 69.65 Jarque-Bera (JB): 6422892.15
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 12.42
Prob(H) (two-sided): 0.00 Kurtosis: 439.89
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 2.74e+29. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
WARNING:tensorflow:Layer lstm_52 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
WARNING:tensorflow:Layer lstm_52 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.84951, saving model to LSTM5.h5 10/10 - 2s - loss: 0.2793 - val_loss: 0.8495 - lr: 0.0010 - 2s/epoch - 166ms/step Epoch 2/500 Epoch 00002: val_loss improved from 0.84951 to 0.03753, saving model to LSTM5.h5 10/10 - 0s - loss: 0.0845 - val_loss: 0.0375 - lr: 0.0010 - 150ms/epoch - 15ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0633 - val_loss: 0.1873 - lr: 0.0010 - 146ms/epoch - 15ms/step Epoch 4/500 Epoch 00004: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0523 - val_loss: 0.4476 - lr: 0.0010 - 111ms/epoch - 11ms/step Epoch 5/500 Epoch 00005: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0470 - val_loss: 0.3455 - lr: 0.0010 - 108ms/epoch - 11ms/step Epoch 6/500 Epoch 00006: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0551 - val_loss: 0.1638 - lr: 0.0010 - 109ms/epoch - 11ms/step Epoch 7/500 Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00007: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0432 - val_loss: 0.1560 - lr: 0.0010 - 104ms/epoch - 10ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0419 - val_loss: 0.1590 - lr: 1.0000e-04 - 92ms/epoch - 9ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0365 - val_loss: 0.1727 - lr: 1.0000e-04 - 97ms/epoch - 10ms/step Epoch 10/500 Epoch 00010: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0386 - val_loss: 0.1835 - lr: 1.0000e-04 - 132ms/epoch - 13ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0384 - val_loss: 0.1881 - lr: 1.0000e-04 - 100ms/epoch - 10ms/step Epoch 12/500 Epoch 00012: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00012: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0342 - val_loss: 0.2006 - lr: 1.0000e-04 - 117ms/epoch - 12ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0372 - val_loss: 0.2009 - lr: 1.0000e-05 - 95ms/epoch - 9ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0342 - val_loss: 0.2012 - lr: 1.0000e-05 - 101ms/epoch - 10ms/step Epoch 15/500 Epoch 00015: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0348 - val_loss: 0.2015 - lr: 1.0000e-05 - 116ms/epoch - 12ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0374 - val_loss: 0.2014 - lr: 1.0000e-05 - 111ms/epoch - 11ms/step Epoch 17/500 Epoch 00017: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00017: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0333 - val_loss: 0.2012 - lr: 1.0000e-05 - 99ms/epoch - 10ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0353 - val_loss: 0.2006 - lr: 1.0000e-05 - 101ms/epoch - 10ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0338 - val_loss: 0.1999 - lr: 1.0000e-05 - 102ms/epoch - 10ms/step Epoch 20/500 Epoch 00020: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0348 - val_loss: 0.1989 - lr: 1.0000e-05 - 98ms/epoch - 10ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0348 - val_loss: 0.1979 - lr: 1.0000e-05 - 95ms/epoch - 9ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0390 - val_loss: 0.1979 - lr: 1.0000e-05 - 89ms/epoch - 9ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0363 - val_loss: 0.1990 - lr: 1.0000e-05 - 106ms/epoch - 11ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0354 - val_loss: 0.1984 - lr: 1.0000e-05 - 88ms/epoch - 9ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0327 - val_loss: 0.1976 - lr: 1.0000e-05 - 107ms/epoch - 11ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0339 - val_loss: 0.1970 - lr: 1.0000e-05 - 96ms/epoch - 10ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0350 - val_loss: 0.1959 - lr: 1.0000e-05 - 106ms/epoch - 11ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0373 - val_loss: 0.1950 - lr: 1.0000e-05 - 91ms/epoch - 9ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0361 - val_loss: 0.1931 - lr: 1.0000e-05 - 97ms/epoch - 10ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0352 - val_loss: 0.1914 - lr: 1.0000e-05 - 107ms/epoch - 11ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0342 - val_loss: 0.1902 - lr: 1.0000e-05 - 104ms/epoch - 10ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0344 - val_loss: 0.1885 - lr: 1.0000e-05 - 91ms/epoch - 9ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0362 - val_loss: 0.1871 - lr: 1.0000e-05 - 99ms/epoch - 10ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0352 - val_loss: 0.1860 - lr: 1.0000e-05 - 96ms/epoch - 10ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0355 - val_loss: 0.1847 - lr: 1.0000e-05 - 128ms/epoch - 13ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0363 - val_loss: 0.1846 - lr: 1.0000e-05 - 109ms/epoch - 11ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0348 - val_loss: 0.1840 - lr: 1.0000e-05 - 98ms/epoch - 10ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0353 - val_loss: 0.1837 - lr: 1.0000e-05 - 116ms/epoch - 12ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0367 - val_loss: 0.1833 - lr: 1.0000e-05 - 99ms/epoch - 10ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0352 - val_loss: 0.1834 - lr: 1.0000e-05 - 104ms/epoch - 10ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0345 - val_loss: 0.1835 - lr: 1.0000e-05 - 96ms/epoch - 10ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0354 - val_loss: 0.1850 - lr: 1.0000e-05 - 111ms/epoch - 11ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0361 - val_loss: 0.1866 - lr: 1.0000e-05 - 110ms/epoch - 11ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0341 - val_loss: 0.1868 - lr: 1.0000e-05 - 112ms/epoch - 11ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0345 - val_loss: 0.1869 - lr: 1.0000e-05 - 100ms/epoch - 10ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0343 - val_loss: 0.1881 - lr: 1.0000e-05 - 109ms/epoch - 11ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0363 - val_loss: 0.1892 - lr: 1.0000e-05 - 109ms/epoch - 11ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0347 - val_loss: 0.1887 - lr: 1.0000e-05 - 104ms/epoch - 10ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0351 - val_loss: 0.1883 - lr: 1.0000e-05 - 98ms/epoch - 10ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0342 - val_loss: 0.1872 - lr: 1.0000e-05 - 90ms/epoch - 9ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0385 - val_loss: 0.1874 - lr: 1.0000e-05 - 112ms/epoch - 11ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.03753 10/10 - 0s - loss: 0.0331 - val_loss: 0.1878 - lr: 1.0000e-05 - 100ms/epoch - 10ms/step Epoch 00052: early stopping
SMA
Prediction vs Close: 46.27% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 22.905457367129987
RMSE: 4.785964622427749
MAPE: 4.003866138329542
EMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 44.03% Accuracy
MSE: 24.329194491765858
RMSE: 4.932463328983385
MAPE: 4.099346553838579
WMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 45.15% Accuracy
MSE: 26.731339639305343
RMSE: 5.170235936522176
MAPE: 4.142288801040536
DEMA
Prediction vs Close: 48.13% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 20.555600166870196
RMSE: 4.53382842274277
MAPE: 3.6522177332314283
KAMA
KAMA([input_arrays], [timeperiod=30])
Kaufman Adaptive Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
18
Working on KAMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16921.943, Time=10.75 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.592, Time=4.54 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-16797.275, Time=9.33 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.592, Time=7.05 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16996.465, Time=3.51 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=-16999.509, Time=3.29 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=-17171.315, Time=6.52 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=-16994.523, Time=3.87 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=-15518.026, Time=29.54 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 78.397 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood 8613.658
Date: Sun, 12 Dec 2021 AIC -17171.315
Time: 15:37:20 BIC -17039.972
Sample: 0 HQIC -17120.874
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -5.14e-10 7.6e-05 -6.76e-06 1.000 -0.000 0.000
x2 -5.041e-10 7.52e-05 -6.7e-06 1.000 -0.000 0.000
x3 -4.834e-10 7.38e-05 -6.55e-06 1.000 -0.000 0.000
x4 1.0000 7.46e-05 1.34e+04 0.000 1.000 1.000
x5 -4.462e-10 7.09e-05 -6.29e-06 1.000 -0.000 0.000
x6 -3.064e-09 0.000 -1.84e-05 1.000 -0.000 0.000
x7 -4.751e-10 7.35e-05 -6.46e-06 1.000 -0.000 0.000
x8 -4.628e-10 7.28e-05 -6.36e-06 1.000 -0.000 0.000
x9 -9.21e-11 9.37e-06 -9.83e-06 1.000 -1.84e-05 1.84e-05
x10 -2.165e-10 3.1e-05 -6.98e-06 1.000 -6.08e-05 6.08e-05
x11 -4.665e-10 7.28e-05 -6.41e-06 1.000 -0.000 0.000
x12 -4.62e-10 7.23e-05 -6.39e-06 1.000 -0.000 0.000
x13 -4.906e-10 7.43e-05 -6.6e-06 1.000 -0.000 0.000
x14 -3.985e-09 0.000 -1.87e-05 1.000 -0.000 0.000
x15 -4.897e-10 7.48e-05 -6.55e-06 1.000 -0.000 0.000
x16 -7.327e-10 9.24e-05 -7.93e-06 1.000 -0.000 0.000
x17 -4.173e-10 6.93e-05 -6.02e-06 1.000 -0.000 0.000
x18 -3.397e-10 6.02e-05 -5.64e-06 1.000 -0.000 0.000
x19 -6.012e-10 8.3e-05 -7.25e-06 1.000 -0.000 0.000
x20 -9.09e-10 0.000 -9.05e-06 1.000 -0.000 0.000
x21 -6.188e-09 0.000 -2.32e-05 1.000 -0.001 0.001
x22 -1.992e-09 0.000 -1.33e-05 1.000 -0.000 0.000
x23 -3.669e-09 0.000 -1.79e-05 1.000 -0.000 0.000
x24 -1.065e-09 0.000 -1.01e-05 1.000 -0.000 0.000
ar.L1 -1.2073 5.73e-10 -2.11e+09 0.000 -1.207 -1.207
ar.L2 -0.9083 5.93e-10 -1.53e+09 0.000 -0.908 -0.908
ar.L3 -0.4033 5.84e-10 -6.91e+08 0.000 -0.403 -0.403
sigma2 8.06e-11 6.94e-11 1.162 0.245 -5.54e-11 2.17e-10
===================================================================================
Ljung-Box (L1) (Q): 13.77 Jarque-Bera (JB): 2436796.68
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 4.07
Prob(H) (two-sided): 0.00 Kurtosis: 272.41
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.23e+28. Standard errors may be unstable.
ARIMA order: (3, 3, 0)
WARNING:tensorflow:Layer lstm_53 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
WARNING:tensorflow:Layer lstm_53 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.31708, saving model to LSTM5.h5 45/45 - 2s - loss: 0.1855 - val_loss: 0.3171 - lr: 0.0010 - 2s/epoch - 44ms/step Epoch 2/500 Epoch 00002: val_loss did not improve from 0.31708 45/45 - 0s - loss: 0.1562 - val_loss: 0.7818 - lr: 0.0010 - 363ms/epoch - 8ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.31708 45/45 - 0s - loss: 0.0779 - val_loss: 0.5710 - lr: 0.0010 - 384ms/epoch - 9ms/step Epoch 4/500 Epoch 00004: val_loss improved from 0.31708 to 0.21579, saving model to LSTM5.h5 45/45 - 0s - loss: 0.0597 - val_loss: 0.2158 - lr: 0.0010 - 416ms/epoch - 9ms/step Epoch 5/500 Epoch 00005: val_loss improved from 0.21579 to 0.13465, saving model to LSTM5.h5 45/45 - 0s - loss: 0.0558 - val_loss: 0.1346 - lr: 0.0010 - 449ms/epoch - 10ms/step Epoch 6/500 Epoch 00006: val_loss improved from 0.13465 to 0.10282, saving model to LSTM5.h5 45/45 - 0s - loss: 0.0428 - val_loss: 0.1028 - lr: 0.0010 - 434ms/epoch - 10ms/step Epoch 7/500 Epoch 00007: val_loss improved from 0.10282 to 0.09417, saving model to LSTM5.h5 45/45 - 0s - loss: 0.0417 - val_loss: 0.0942 - lr: 0.0010 - 419ms/epoch - 9ms/step Epoch 8/500 Epoch 00008: val_loss improved from 0.09417 to 0.06728, saving model to LSTM5.h5 45/45 - 0s - loss: 0.0325 - val_loss: 0.0673 - lr: 0.0010 - 419ms/epoch - 9ms/step Epoch 9/500 Epoch 00009: val_loss improved from 0.06728 to 0.02801, saving model to LSTM5.h5 45/45 - 0s - loss: 0.0371 - val_loss: 0.0280 - lr: 0.0010 - 372ms/epoch - 8ms/step Epoch 10/500 Epoch 00010: val_loss did not improve from 0.02801 45/45 - 0s - loss: 0.0375 - val_loss: 0.1020 - lr: 0.0010 - 390ms/epoch - 9ms/step Epoch 11/500 Epoch 00011: val_loss improved from 0.02801 to 0.02044, saving model to LSTM5.h5 45/45 - 0s - loss: 0.0345 - val_loss: 0.0204 - lr: 0.0010 - 377ms/epoch - 8ms/step Epoch 12/500 Epoch 00012: val_loss did not improve from 0.02044 45/45 - 0s - loss: 0.0361 - val_loss: 0.1002 - lr: 0.0010 - 370ms/epoch - 8ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.02044 45/45 - 0s - loss: 0.0303 - val_loss: 0.0225 - lr: 0.0010 - 361ms/epoch - 8ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.02044 45/45 - 0s - loss: 0.0317 - val_loss: 0.0947 - lr: 0.0010 - 368ms/epoch - 8ms/step Epoch 15/500 Epoch 00015: val_loss improved from 0.02044 to 0.01486, saving model to LSTM5.h5 45/45 - 0s - loss: 0.0339 - val_loss: 0.0149 - lr: 0.0010 - 378ms/epoch - 8ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.01486 45/45 - 0s - loss: 0.0341 - val_loss: 0.1791 - lr: 0.0010 - 363ms/epoch - 8ms/step Epoch 17/500 Epoch 00017: val_loss did not improve from 0.01486 45/45 - 0s - loss: 0.0316 - val_loss: 0.0240 - lr: 0.0010 - 354ms/epoch - 8ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.01486 45/45 - 0s - loss: 0.0266 - val_loss: 0.0898 - lr: 0.0010 - 386ms/epoch - 9ms/step Epoch 19/500 Epoch 00019: val_loss improved from 0.01486 to 0.01280, saving model to LSTM5.h5 45/45 - 0s - loss: 0.0376 - val_loss: 0.0128 - lr: 0.0010 - 423ms/epoch - 9ms/step Epoch 20/500 Epoch 00020: val_loss did not improve from 0.01280 45/45 - 0s - loss: 0.0332 - val_loss: 0.0765 - lr: 0.0010 - 363ms/epoch - 8ms/step Epoch 21/500 Epoch 00021: val_loss improved from 0.01280 to 0.01098, saving model to LSTM5.h5 45/45 - 0s - loss: 0.0252 - val_loss: 0.0110 - lr: 0.0010 - 412ms/epoch - 9ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.01098 45/45 - 0s - loss: 0.0326 - val_loss: 0.1760 - lr: 0.0010 - 414ms/epoch - 9ms/step Epoch 23/500 Epoch 00023: val_loss improved from 0.01098 to 0.01031, saving model to LSTM5.h5 45/45 - 0s - loss: 0.0246 - val_loss: 0.0103 - lr: 0.0010 - 487ms/epoch - 11ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0223 - val_loss: 0.1079 - lr: 0.0010 - 361ms/epoch - 8ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0265 - val_loss: 0.0105 - lr: 0.0010 - 357ms/epoch - 8ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0224 - val_loss: 0.0331 - lr: 0.0010 - 347ms/epoch - 8ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0242 - val_loss: 0.0119 - lr: 0.0010 - 386ms/epoch - 9ms/step Epoch 28/500 Epoch 00028: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00028: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0233 - val_loss: 0.1756 - lr: 0.0010 - 376ms/epoch - 8ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0292 - val_loss: 0.1420 - lr: 1.0000e-04 - 377ms/epoch - 8ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0205 - val_loss: 0.1177 - lr: 1.0000e-04 - 388ms/epoch - 9ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0219 - val_loss: 0.0977 - lr: 1.0000e-04 - 447ms/epoch - 10ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0193 - val_loss: 0.0782 - lr: 1.0000e-04 - 386ms/epoch - 9ms/step Epoch 33/500 Epoch 00033: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00033: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0192 - val_loss: 0.0651 - lr: 1.0000e-04 - 383ms/epoch - 9ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0191 - val_loss: 0.0642 - lr: 1.0000e-05 - 374ms/epoch - 8ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0196 - val_loss: 0.0634 - lr: 1.0000e-05 - 422ms/epoch - 9ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0191 - val_loss: 0.0627 - lr: 1.0000e-05 - 415ms/epoch - 9ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0186 - val_loss: 0.0618 - lr: 1.0000e-05 - 363ms/epoch - 8ms/step Epoch 38/500 Epoch 00038: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00038: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0218 - val_loss: 0.0608 - lr: 1.0000e-05 - 360ms/epoch - 8ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0191 - val_loss: 0.0598 - lr: 1.0000e-05 - 353ms/epoch - 8ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0191 - val_loss: 0.0590 - lr: 1.0000e-05 - 366ms/epoch - 8ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0180 - val_loss: 0.0581 - lr: 1.0000e-05 - 352ms/epoch - 8ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0180 - val_loss: 0.0576 - lr: 1.0000e-05 - 357ms/epoch - 8ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0196 - val_loss: 0.0563 - lr: 1.0000e-05 - 386ms/epoch - 9ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0192 - val_loss: 0.0555 - lr: 1.0000e-05 - 409ms/epoch - 9ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0195 - val_loss: 0.0549 - lr: 1.0000e-05 - 354ms/epoch - 8ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0193 - val_loss: 0.0538 - lr: 1.0000e-05 - 378ms/epoch - 8ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0186 - val_loss: 0.0530 - lr: 1.0000e-05 - 379ms/epoch - 8ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0185 - val_loss: 0.0524 - lr: 1.0000e-05 - 407ms/epoch - 9ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0174 - val_loss: 0.0518 - lr: 1.0000e-05 - 394ms/epoch - 9ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0199 - val_loss: 0.0510 - lr: 1.0000e-05 - 405ms/epoch - 9ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0185 - val_loss: 0.0505 - lr: 1.0000e-05 - 403ms/epoch - 9ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0204 - val_loss: 0.0495 - lr: 1.0000e-05 - 394ms/epoch - 9ms/step Epoch 53/500 Epoch 00053: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0184 - val_loss: 0.0491 - lr: 1.0000e-05 - 371ms/epoch - 8ms/step Epoch 54/500 Epoch 00054: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0198 - val_loss: 0.0492 - lr: 1.0000e-05 - 373ms/epoch - 8ms/step Epoch 55/500 Epoch 00055: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0187 - val_loss: 0.0480 - lr: 1.0000e-05 - 347ms/epoch - 8ms/step Epoch 56/500 Epoch 00056: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0191 - val_loss: 0.0471 - lr: 1.0000e-05 - 395ms/epoch - 9ms/step Epoch 57/500 Epoch 00057: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0175 - val_loss: 0.0460 - lr: 1.0000e-05 - 395ms/epoch - 9ms/step Epoch 58/500 Epoch 00058: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0165 - val_loss: 0.0452 - lr: 1.0000e-05 - 415ms/epoch - 9ms/step Epoch 59/500 Epoch 00059: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0162 - val_loss: 0.0443 - lr: 1.0000e-05 - 416ms/epoch - 9ms/step Epoch 60/500 Epoch 00060: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0174 - val_loss: 0.0435 - lr: 1.0000e-05 - 353ms/epoch - 8ms/step Epoch 61/500 Epoch 00061: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0176 - val_loss: 0.0428 - lr: 1.0000e-05 - 409ms/epoch - 9ms/step Epoch 62/500 Epoch 00062: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0178 - val_loss: 0.0423 - lr: 1.0000e-05 - 376ms/epoch - 8ms/step Epoch 63/500 Epoch 00063: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0187 - val_loss: 0.0412 - lr: 1.0000e-05 - 387ms/epoch - 9ms/step Epoch 64/500 Epoch 00064: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0179 - val_loss: 0.0408 - lr: 1.0000e-05 - 399ms/epoch - 9ms/step Epoch 65/500 Epoch 00065: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0180 - val_loss: 0.0405 - lr: 1.0000e-05 - 354ms/epoch - 8ms/step Epoch 66/500 Epoch 00066: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0176 - val_loss: 0.0395 - lr: 1.0000e-05 - 360ms/epoch - 8ms/step Epoch 67/500 Epoch 00067: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0177 - val_loss: 0.0394 - lr: 1.0000e-05 - 346ms/epoch - 8ms/step Epoch 68/500 Epoch 00068: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0183 - val_loss: 0.0387 - lr: 1.0000e-05 - 381ms/epoch - 8ms/step Epoch 69/500 Epoch 00069: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0171 - val_loss: 0.0381 - lr: 1.0000e-05 - 397ms/epoch - 9ms/step Epoch 70/500 Epoch 00070: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0180 - val_loss: 0.0370 - lr: 1.0000e-05 - 417ms/epoch - 9ms/step Epoch 71/500 Epoch 00071: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0200 - val_loss: 0.0374 - lr: 1.0000e-05 - 364ms/epoch - 8ms/step Epoch 72/500 Epoch 00072: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0188 - val_loss: 0.0366 - lr: 1.0000e-05 - 361ms/epoch - 8ms/step Epoch 73/500 Epoch 00073: val_loss did not improve from 0.01031 45/45 - 0s - loss: 0.0202 - val_loss: 0.0367 - lr: 1.0000e-05 - 348ms/epoch - 8ms/step Epoch 00073: early stopping
SMA
Prediction vs Close: 46.27% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 22.905457367129987
RMSE: 4.785964622427749
MAPE: 4.003866138329542
EMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 44.03% Accuracy
MSE: 24.329194491765858
RMSE: 4.932463328983385
MAPE: 4.099346553838579
WMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 45.15% Accuracy
MSE: 26.731339639305343
RMSE: 5.170235936522176
MAPE: 4.142288801040536
DEMA
Prediction vs Close: 48.13% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 20.555600166870196
RMSE: 4.53382842274277
MAPE: 3.6522177332314283
KAMA
Prediction vs Close: 57.84% Accuracy
Prediction vs Prediction: 44.78% Accuracy
MSE: 36.658985139129086
RMSE: 6.054666393710646
MAPE: 4.91375972579294
MIDPOINT
MIDPOINT([input_arrays], [timeperiod=14])
MidPoint over period (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 14
Outputs:
real
14
Working on MIDPOINT predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.768, Time=3.28 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.591, Time=4.56 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-15581.065, Time=8.79 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.591, Time=7.49 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16536.628, Time=9.58 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-13971.493, Time=10.43 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17226.044, Time=21.15 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-14562.591, Time=9.27 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-16754.945, Time=19.51 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-15001.855, Time=21.51 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 115.600 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8640.022
Date: Sun, 12 Dec 2021 AIC -17226.044
Time: 15:41:32 BIC -17099.391
Sample: 0 HQIC -17177.404
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -5.031e-09 1.06e-05 -0.000 1.000 -2.08e-05 2.08e-05
x2 -4.99e-09 8.12e-06 -0.001 1.000 -1.59e-05 1.59e-05
x3 -5.114e-09 1.38e-05 -0.000 1.000 -2.7e-05 2.7e-05
x4 1.0000 8.91e-06 1.12e+05 0.000 1.000 1.000
x5 -4.55e-09 8.2e-06 -0.001 1.000 -1.61e-05 1.61e-05
x6 -9.992e-08 0.001 -0.000 1.000 -0.002 0.002
x7 -4.607e-09 1.97e-05 -0.000 1.000 -3.86e-05 3.86e-05
x8 -4.591e-09 1.77e-05 -0.000 1.000 -3.48e-05 3.48e-05
x9 -2.538e-09 1.13e-05 -0.000 1.000 -2.21e-05 2.21e-05
x10 -4.315e-09 6.08e-06 -0.001 0.999 -1.19e-05 1.19e-05
x11 -4.545e-09 1.62e-05 -0.000 1.000 -3.18e-05 3.18e-05
x12 -4.701e-09 1.97e-05 -0.000 1.000 -3.87e-05 3.87e-05
x13 -4.823e-09 1.18e-05 -0.000 1.000 -2.3e-05 2.3e-05
x14 -4.08e-08 4.99e-05 -0.001 0.999 -9.79e-05 9.78e-05
x15 -5.557e-09 2.03e-05 -0.000 1.000 -3.99e-05 3.99e-05
x16 -3.541e-09 1.3e-05 -0.000 1.000 -2.55e-05 2.55e-05
x17 -3.463e-09 1.51e-05 -0.000 1.000 -2.97e-05 2.97e-05
x18 -1.534e-08 4e-05 -0.000 1.000 -7.85e-05 7.85e-05
x19 -6.118e-09 2.07e-05 -0.000 1.000 -4.05e-05 4.05e-05
x20 -1.581e-08 3.38e-05 -0.000 1.000 -6.62e-05 6.61e-05
x21 -5.505e-08 5.6e-05 -0.001 0.999 -0.000 0.000
x22 -2.936e-08 4.55e-05 -0.001 0.999 -8.92e-05 8.92e-05
x23 -3.882e-08 4.89e-05 -0.001 0.999 -9.58e-05 9.57e-05
x24 -2.099e-08 4.87e-05 -0.000 1.000 -9.54e-05 9.54e-05
ma.L1 -1.3900 1.23e-07 -1.13e+07 0.000 -1.390 -1.390
ma.L2 0.4044 1.43e-07 2.82e+06 0.000 0.404 0.404
sigma2 7.525e-11 7.22e-11 1.042 0.297 -6.63e-11 2.17e-10
===================================================================================
Ljung-Box (L1) (Q): 55.84 Jarque-Bera (JB): 1335305.59
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.09 Skew: 5.74
Prob(H) (two-sided): 0.00 Kurtosis: 202.19
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.77e+23. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
WARNING:tensorflow:Layer lstm_54 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
WARNING:tensorflow:Layer lstm_54 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.10748, saving model to LSTM5.h5 58/58 - 2s - loss: 0.1689 - val_loss: 0.1075 - lr: 0.0010 - 2s/epoch - 37ms/step Epoch 2/500 Epoch 00002: val_loss did not improve from 0.10748 58/58 - 0s - loss: 0.1680 - val_loss: 0.1948 - lr: 0.0010 - 499ms/epoch - 9ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.10748 58/58 - 0s - loss: 0.0806 - val_loss: 0.2118 - lr: 0.0010 - 480ms/epoch - 8ms/step Epoch 4/500 Epoch 00004: val_loss improved from 0.10748 to 0.02087, saving model to LSTM5.h5 58/58 - 0s - loss: 0.0585 - val_loss: 0.0209 - lr: 0.0010 - 491ms/epoch - 8ms/step Epoch 5/500 Epoch 00005: val_loss did not improve from 0.02087 58/58 - 0s - loss: 0.0409 - val_loss: 0.1412 - lr: 0.0010 - 480ms/epoch - 8ms/step Epoch 6/500 Epoch 00006: val_loss improved from 0.02087 to 0.00898, saving model to LSTM5.h5 58/58 - 1s - loss: 0.0370 - val_loss: 0.0090 - lr: 0.0010 - 515ms/epoch - 9ms/step Epoch 7/500 Epoch 00007: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0341 - val_loss: 0.0605 - lr: 0.0010 - 489ms/epoch - 8ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0395 - val_loss: 0.0121 - lr: 0.0010 - 491ms/epoch - 8ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0363 - val_loss: 0.1334 - lr: 0.0010 - 478ms/epoch - 8ms/step Epoch 10/500 Epoch 00010: val_loss did not improve from 0.00898 58/58 - 1s - loss: 0.0293 - val_loss: 0.0147 - lr: 0.0010 - 538ms/epoch - 9ms/step Epoch 11/500 Epoch 00011: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00011: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0259 - val_loss: 0.0226 - lr: 0.0010 - 474ms/epoch - 8ms/step Epoch 12/500 Epoch 00012: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0300 - val_loss: 0.0235 - lr: 1.0000e-04 - 499ms/epoch - 9ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0270 - val_loss: 0.0340 - lr: 1.0000e-04 - 482ms/epoch - 8ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0257 - val_loss: 0.0341 - lr: 1.0000e-04 - 466ms/epoch - 8ms/step Epoch 15/500 Epoch 00015: val_loss did not improve from 0.00898 58/58 - 1s - loss: 0.0250 - val_loss: 0.0326 - lr: 1.0000e-04 - 500ms/epoch - 9ms/step Epoch 16/500 Epoch 00016: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00016: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0234 - val_loss: 0.0340 - lr: 1.0000e-04 - 475ms/epoch - 8ms/step Epoch 17/500 Epoch 00017: val_loss did not improve from 0.00898 58/58 - 1s - loss: 0.0251 - val_loss: 0.0342 - lr: 1.0000e-05 - 508ms/epoch - 9ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0253 - val_loss: 0.0340 - lr: 1.0000e-05 - 470ms/epoch - 8ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0248 - val_loss: 0.0336 - lr: 1.0000e-05 - 485ms/epoch - 8ms/step Epoch 20/500 Epoch 00020: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0236 - val_loss: 0.0334 - lr: 1.0000e-05 - 475ms/epoch - 8ms/step Epoch 21/500 Epoch 00021: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00021: val_loss did not improve from 0.00898 58/58 - 1s - loss: 0.0269 - val_loss: 0.0328 - lr: 1.0000e-05 - 508ms/epoch - 9ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0228 - val_loss: 0.0323 - lr: 1.0000e-05 - 483ms/epoch - 8ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.00898 58/58 - 1s - loss: 0.0249 - val_loss: 0.0321 - lr: 1.0000e-05 - 553ms/epoch - 10ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.00898 58/58 - 1s - loss: 0.0230 - val_loss: 0.0314 - lr: 1.0000e-05 - 504ms/epoch - 9ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.00898 58/58 - 1s - loss: 0.0241 - val_loss: 0.0317 - lr: 1.0000e-05 - 551ms/epoch - 9ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0241 - val_loss: 0.0319 - lr: 1.0000e-05 - 486ms/epoch - 8ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0215 - val_loss: 0.0318 - lr: 1.0000e-05 - 488ms/epoch - 8ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.00898 58/58 - 1s - loss: 0.0245 - val_loss: 0.0321 - lr: 1.0000e-05 - 509ms/epoch - 9ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0258 - val_loss: 0.0326 - lr: 1.0000e-05 - 458ms/epoch - 8ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0252 - val_loss: 0.0324 - lr: 1.0000e-05 - 473ms/epoch - 8ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0237 - val_loss: 0.0323 - lr: 1.0000e-05 - 498ms/epoch - 9ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0245 - val_loss: 0.0316 - lr: 1.0000e-05 - 477ms/epoch - 8ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0229 - val_loss: 0.0312 - lr: 1.0000e-05 - 458ms/epoch - 8ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0234 - val_loss: 0.0306 - lr: 1.0000e-05 - 492ms/epoch - 8ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.00898 58/58 - 1s - loss: 0.0240 - val_loss: 0.0302 - lr: 1.0000e-05 - 508ms/epoch - 9ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0223 - val_loss: 0.0297 - lr: 1.0000e-05 - 466ms/epoch - 8ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.00898 58/58 - 1s - loss: 0.0262 - val_loss: 0.0299 - lr: 1.0000e-05 - 521ms/epoch - 9ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0258 - val_loss: 0.0293 - lr: 1.0000e-05 - 462ms/epoch - 8ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0211 - val_loss: 0.0285 - lr: 1.0000e-05 - 469ms/epoch - 8ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0240 - val_loss: 0.0282 - lr: 1.0000e-05 - 487ms/epoch - 8ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.00898 58/58 - 1s - loss: 0.0240 - val_loss: 0.0273 - lr: 1.0000e-05 - 513ms/epoch - 9ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0247 - val_loss: 0.0267 - lr: 1.0000e-05 - 495ms/epoch - 9ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0217 - val_loss: 0.0269 - lr: 1.0000e-05 - 485ms/epoch - 8ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0232 - val_loss: 0.0270 - lr: 1.0000e-05 - 497ms/epoch - 9ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0220 - val_loss: 0.0272 - lr: 1.0000e-05 - 494ms/epoch - 9ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0238 - val_loss: 0.0279 - lr: 1.0000e-05 - 471ms/epoch - 8ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0239 - val_loss: 0.0276 - lr: 1.0000e-05 - 499ms/epoch - 9ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0232 - val_loss: 0.0266 - lr: 1.0000e-05 - 466ms/epoch - 8ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0230 - val_loss: 0.0257 - lr: 1.0000e-05 - 498ms/epoch - 9ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0224 - val_loss: 0.0248 - lr: 1.0000e-05 - 462ms/epoch - 8ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0226 - val_loss: 0.0254 - lr: 1.0000e-05 - 485ms/epoch - 8ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.00898 58/58 - 1s - loss: 0.0212 - val_loss: 0.0269 - lr: 1.0000e-05 - 518ms/epoch - 9ms/step Epoch 53/500 Epoch 00053: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0217 - val_loss: 0.0270 - lr: 1.0000e-05 - 479ms/epoch - 8ms/step Epoch 54/500 Epoch 00054: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0233 - val_loss: 0.0266 - lr: 1.0000e-05 - 458ms/epoch - 8ms/step Epoch 55/500 Epoch 00055: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0240 - val_loss: 0.0280 - lr: 1.0000e-05 - 488ms/epoch - 8ms/step Epoch 56/500 Epoch 00056: val_loss did not improve from 0.00898 58/58 - 0s - loss: 0.0221 - val_loss: 0.0271 - lr: 1.0000e-05 - 475ms/epoch - 8ms/step Epoch 00056: early stopping
SMA
Prediction vs Close: 46.27% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 22.905457367129987
RMSE: 4.785964622427749
MAPE: 4.003866138329542
EMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 44.03% Accuracy
MSE: 24.329194491765858
RMSE: 4.932463328983385
MAPE: 4.099346553838579
WMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 45.15% Accuracy
MSE: 26.731339639305343
RMSE: 5.170235936522176
MAPE: 4.142288801040536
DEMA
Prediction vs Close: 48.13% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 20.555600166870196
RMSE: 4.53382842274277
MAPE: 3.6522177332314283
KAMA
Prediction vs Close: 57.84% Accuracy
Prediction vs Prediction: 44.78% Accuracy
MSE: 36.658985139129086
RMSE: 6.054666393710646
MAPE: 4.91375972579294
MIDPOINT
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 44.4% Accuracy
MSE: 46.211728483578064
RMSE: 6.797920894183608
MAPE: 5.510818514624332
T3
T3([input_arrays], [timeperiod=5], [vfactor=0.7])
Triple Exponential Moving Average (T3) (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 5
vfactor: 0.7
Outputs:
real
19
Working on T3 predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-17000.569, Time=3.24 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-15576.554, Time=5.69 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-16078.305, Time=8.02 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-15574.554, Time=8.99 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16998.627, Time=3.52 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-16429.916, Time=12.84 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17000.664, Time=3.56 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-15700.026, Time=11.86 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-15704.282, Time=15.01 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-16998.664, Time=3.14 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 75.889 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8527.332
Date: Sun, 12 Dec 2021 AIC -17000.664
Time: 15:47:27 BIC -16874.011
Sample: 0 HQIC -16952.024
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 8.378e-14 2.16e-06 3.89e-08 1.000 -4.23e-06 4.23e-06
x2 7.457e-14 2.15e-06 3.47e-08 1.000 -4.22e-06 4.22e-06
x3 2.279e-14 2.16e-06 1.05e-08 1.000 -4.24e-06 4.24e-06
x4 1.0000 2.16e-06 4.63e+05 0.000 1.000 1.000
x5 1.211e-12 2.07e-06 5.86e-07 1.000 -4.05e-06 4.05e-06
x6 3.146e-15 2.67e-06 1.18e-09 1.000 -5.23e-06 5.23e-06
x7 1.593e-13 2.15e-06 7.41e-08 1.000 -4.21e-06 4.21e-06
x8 -0.0001 2.1e-06 -48.778 0.000 -0.000 -9.82e-05
x9 5.141e-14 6.35e-07 8.1e-08 1.000 -1.24e-06 1.24e-06
x10 -6.174e-05 1.34e-06 -45.995 0.000 -6.44e-05 -5.91e-05
x11 0.0003 2.15e-06 148.354 0.000 0.000 0.000
x12 -0.0002 2.02e-06 -93.730 0.000 -0.000 -0.000
x13 1.967e-14 2.16e-06 9.1e-09 1.000 -4.23e-06 4.23e-06
x14 -1.297e-14 5.65e-06 -2.29e-09 1.000 -1.11e-05 1.11e-05
x15 -3.18e-12 1.82e-06 -1.75e-06 1.000 -3.57e-06 3.57e-06
x16 -1.426e-12 4.51e-06 -3.16e-07 1.000 -8.84e-06 8.84e-06
x17 7.474e-13 2.37e-06 3.16e-07 1.000 -4.64e-06 4.64e-06
x18 -2.92e-13 2.9e-06 -1.01e-07 1.000 -5.68e-06 5.68e-06
x19 -4.211e-14 1.89e-06 -2.22e-08 1.000 -3.71e-06 3.71e-06
x20 -1.515e-13 1.2e-06 -1.26e-07 1.000 -2.36e-06 2.36e-06
x21 6.555e-13 6.37e-06 1.03e-07 1.000 -1.25e-05 1.25e-05
x22 1.212e-14 6.19e-06 1.96e-09 1.000 -1.21e-05 1.21e-05
x23 -3.877e-13 3.76e-06 -1.03e-07 1.000 -7.38e-06 7.38e-06
x24 8.127e-15 4.01e-06 2.03e-09 1.000 -7.86e-06 7.86e-06
ma.L1 -1.3370 3.84e-12 -3.48e+11 0.000 -1.337 -1.337
ma.L2 0.4289 1.65e-12 2.6e+11 0.000 0.429 0.429
sigma2 1e-10 6.99e-11 1.430 0.153 -3.71e-11 2.37e-10
===================================================================================
Ljung-Box (L1) (Q): 4.57 Jarque-Bera (JB): 3228712.87
Prob(Q): 0.03 Prob(JB): 0.00
Heteroskedasticity (H): 0.12 Skew: -9.87
Prob(H) (two-sided): 0.00 Kurtosis: 312.63
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.57e+30. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
WARNING:tensorflow:Layer lstm_55 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
WARNING:tensorflow:Layer lstm_55 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.19644, saving model to LSTM5.h5 43/43 - 2s - loss: 0.3641 - val_loss: 0.1964 - lr: 0.0010 - 2s/epoch - 46ms/step Epoch 2/500 Epoch 00002: val_loss improved from 0.19644 to 0.02016, saving model to LSTM5.h5 43/43 - 0s - loss: 0.1791 - val_loss: 0.0202 - lr: 0.0010 - 442ms/epoch - 10ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0733 - val_loss: 0.5417 - lr: 0.0010 - 363ms/epoch - 8ms/step Epoch 4/500 Epoch 00004: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0581 - val_loss: 0.2098 - lr: 0.0010 - 341ms/epoch - 8ms/step Epoch 5/500 Epoch 00005: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0566 - val_loss: 0.0324 - lr: 0.0010 - 405ms/epoch - 9ms/step Epoch 6/500 Epoch 00006: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0494 - val_loss: 0.1509 - lr: 0.0010 - 381ms/epoch - 9ms/step Epoch 7/500 Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00007: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0493 - val_loss: 0.0739 - lr: 0.0010 - 448ms/epoch - 10ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0484 - val_loss: 0.0821 - lr: 1.0000e-04 - 366ms/epoch - 9ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0407 - val_loss: 0.0852 - lr: 1.0000e-04 - 337ms/epoch - 8ms/step Epoch 10/500 Epoch 00010: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0458 - val_loss: 0.0893 - lr: 1.0000e-04 - 356ms/epoch - 8ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0395 - val_loss: 0.0956 - lr: 1.0000e-04 - 370ms/epoch - 9ms/step Epoch 12/500 Epoch 00012: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00012: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0381 - val_loss: 0.0973 - lr: 1.0000e-04 - 380ms/epoch - 9ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0392 - val_loss: 0.0977 - lr: 1.0000e-05 - 373ms/epoch - 9ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0413 - val_loss: 0.0974 - lr: 1.0000e-05 - 431ms/epoch - 10ms/step Epoch 15/500 Epoch 00015: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0393 - val_loss: 0.0967 - lr: 1.0000e-05 - 481ms/epoch - 11ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0398 - val_loss: 0.0960 - lr: 1.0000e-05 - 357ms/epoch - 8ms/step Epoch 17/500 Epoch 00017: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00017: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0374 - val_loss: 0.0955 - lr: 1.0000e-05 - 353ms/epoch - 8ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0380 - val_loss: 0.0952 - lr: 1.0000e-05 - 398ms/epoch - 9ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0404 - val_loss: 0.0959 - lr: 1.0000e-05 - 362ms/epoch - 8ms/step Epoch 20/500 Epoch 00020: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0381 - val_loss: 0.0959 - lr: 1.0000e-05 - 395ms/epoch - 9ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0386 - val_loss: 0.0970 - lr: 1.0000e-05 - 383ms/epoch - 9ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0377 - val_loss: 0.0978 - lr: 1.0000e-05 - 352ms/epoch - 8ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0381 - val_loss: 0.0982 - lr: 1.0000e-05 - 415ms/epoch - 10ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0333 - val_loss: 0.0972 - lr: 1.0000e-05 - 358ms/epoch - 8ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0352 - val_loss: 0.0975 - lr: 1.0000e-05 - 378ms/epoch - 9ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0380 - val_loss: 0.0970 - lr: 1.0000e-05 - 452ms/epoch - 11ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0346 - val_loss: 0.0966 - lr: 1.0000e-05 - 353ms/epoch - 8ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0381 - val_loss: 0.0965 - lr: 1.0000e-05 - 405ms/epoch - 9ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0395 - val_loss: 0.0966 - lr: 1.0000e-05 - 349ms/epoch - 8ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0382 - val_loss: 0.0969 - lr: 1.0000e-05 - 383ms/epoch - 9ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0382 - val_loss: 0.0979 - lr: 1.0000e-05 - 369ms/epoch - 9ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0385 - val_loss: 0.0989 - lr: 1.0000e-05 - 362ms/epoch - 8ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0352 - val_loss: 0.0987 - lr: 1.0000e-05 - 358ms/epoch - 8ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0395 - val_loss: 0.0993 - lr: 1.0000e-05 - 399ms/epoch - 9ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0366 - val_loss: 0.0987 - lr: 1.0000e-05 - 394ms/epoch - 9ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0384 - val_loss: 0.0977 - lr: 1.0000e-05 - 420ms/epoch - 10ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0329 - val_loss: 0.0974 - lr: 1.0000e-05 - 349ms/epoch - 8ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0354 - val_loss: 0.0981 - lr: 1.0000e-05 - 353ms/epoch - 8ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0354 - val_loss: 0.0972 - lr: 1.0000e-05 - 366ms/epoch - 9ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0340 - val_loss: 0.0984 - lr: 1.0000e-05 - 396ms/epoch - 9ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0352 - val_loss: 0.0997 - lr: 1.0000e-05 - 384ms/epoch - 9ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0378 - val_loss: 0.1001 - lr: 1.0000e-05 - 371ms/epoch - 9ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0376 - val_loss: 0.1000 - lr: 1.0000e-05 - 355ms/epoch - 8ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0330 - val_loss: 0.0996 - lr: 1.0000e-05 - 359ms/epoch - 8ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0361 - val_loss: 0.0995 - lr: 1.0000e-05 - 351ms/epoch - 8ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0348 - val_loss: 0.0988 - lr: 1.0000e-05 - 380ms/epoch - 9ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0348 - val_loss: 0.0984 - lr: 1.0000e-05 - 428ms/epoch - 10ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0373 - val_loss: 0.0986 - lr: 1.0000e-05 - 468ms/epoch - 11ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0378 - val_loss: 0.0989 - lr: 1.0000e-05 - 392ms/epoch - 9ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0378 - val_loss: 0.1000 - lr: 1.0000e-05 - 337ms/epoch - 8ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0360 - val_loss: 0.1007 - lr: 1.0000e-05 - 347ms/epoch - 8ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.02016 43/43 - 0s - loss: 0.0372 - val_loss: 0.1004 - lr: 1.0000e-05 - 405ms/epoch - 9ms/step Epoch 00052: early stopping
SMA
Prediction vs Close: 46.27% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 22.905457367129987
RMSE: 4.785964622427749
MAPE: 4.003866138329542
EMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 44.03% Accuracy
MSE: 24.329194491765858
RMSE: 4.932463328983385
MAPE: 4.099346553838579
WMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 45.15% Accuracy
MSE: 26.731339639305343
RMSE: 5.170235936522176
MAPE: 4.142288801040536
DEMA
Prediction vs Close: 48.13% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 20.555600166870196
RMSE: 4.53382842274277
MAPE: 3.6522177332314283
KAMA
Prediction vs Close: 57.84% Accuracy
Prediction vs Prediction: 44.78% Accuracy
MSE: 36.658985139129086
RMSE: 6.054666393710646
MAPE: 4.91375972579294
MIDPOINT
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 44.4% Accuracy
MSE: 46.211728483578064
RMSE: 6.797920894183608
MAPE: 5.510818514624332
T3
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 39.422414094648474
RMSE: 6.278727107833918
MAPE: 5.177910469752962
TEMA
TEMA([input_arrays], [timeperiod=30])
Triple Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
9
Working on TEMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16762.799, Time=5.15 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14158.507, Time=2.90 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-16445.598, Time=9.16 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-16144.282, Time=11.57 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-15275.101, Time=8.94 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-15897.090, Time=14.43 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-16446.973, Time=9.17 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=-16567.628, Time=3.71 sec
/usr/local/lib/python3.7/dist-packages/statsmodels/tsa/statespace/sarimax.py:1906: RuntimeWarning: divide by zero encountered in reciprocal return np.roots(self.polynomial_reduced_ma)**-1
ARIMA(2,3,2)(0,0,0)[0] : AIC=-16523.926, Time=4.41 sec
ARIMA(1,3,1)(0,0,0)[0] intercept : AIC=-16696.008, Time=3.28 sec
Best model: ARIMA(1,3,1)(0,0,0)[0]
Total fit time: 72.746 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(1, 3, 1) Log Likelihood 8408.400
Date: Sun, 12 Dec 2021 AIC -16762.799
Time: 15:53:36 BIC -16636.147
Sample: 0 HQIC -16714.159
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -5.289e-07 0.001 -0.000 1.000 -0.002 0.002
x2 -5.288e-07 0.001 -0.001 0.999 -0.002 0.002
x3 -5.306e-07 0.001 -0.000 1.000 -0.002 0.002
x4 1.0000 0.000 2045.695 0.000 0.999 1.001
x5 -5.041e-07 0.000 -0.001 0.999 -0.001 0.001
x6 -9.879e-07 4.33e-05 -0.023 0.982 -8.58e-05 8.38e-05
x7 -5.185e-07 0.001 -0.001 0.999 -0.001 0.001
x8 0.0001 0.000 0.643 0.520 -0.000 0.001
x9 9.794e-08 0.001 0.000 1.000 -0.001 0.001
x10 0.0001 0.000 0.313 0.754 -0.001 0.001
x11 -0.0004 0.000 -2.284 0.022 -0.001 -6.06e-05
x12 0.0005 0.000 2.453 0.014 0.000 0.001
x13 -5.277e-07 0.000 -0.002 0.999 -0.001 0.001
x14 -1.566e-06 0.000 -0.005 0.996 -0.001 0.001
x15 -5.136e-07 9.86e-05 -0.005 0.996 -0.000 0.000
x16 -7.66e-07 0.000 -0.002 0.999 -0.001 0.001
x17 -5.146e-07 0.000 -0.003 0.998 -0.000 0.000
x18 -1.701e-07 0.001 -0.000 1.000 -0.001 0.001
x19 -5.77e-07 8.54e-05 -0.007 0.995 -0.000 0.000
x20 5.026e-07 0.001 0.001 0.999 -0.001 0.001
x21 -2.058e-06 0.000 -0.010 0.992 -0.000 0.000
x22 -1.098e-06 0.001 -0.001 0.999 -0.003 0.003
x23 -1.472e-06 0.001 -0.003 0.998 -0.001 0.001
x24 -8.255e-07 0.001 -0.001 0.999 -0.002 0.002
ar.L1 -0.2866 3.63e-05 -7897.273 0.000 -0.287 -0.287
ma.L1 -0.9124 1.46e-06 -6.25e+05 0.000 -0.912 -0.912
sigma2 9.98e-11 7.23e-11 1.380 0.168 -4.2e-11 2.42e-10
===================================================================================
Ljung-Box (L1) (Q): 83.51 Jarque-Bera (JB): 4742889.91
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: -5.71
Prob(H) (two-sided): 0.00 Kurtosis: 378.86
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 4.2e+22. Standard errors may be unstable.
ARIMA order: (1, 3, 1)
WARNING:tensorflow:Layer lstm_56 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
WARNING:tensorflow:Layer lstm_56 will not use cuDNN kernels since it doesn't meet the criteria. It will use a generic GPU kernel as fallback when running on GPU.
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.08859, saving model to LSTM5.h5 90/90 - 2s - loss: 0.1608 - val_loss: 0.0886 - lr: 0.0010 - 2s/epoch - 27ms/step Epoch 2/500 Epoch 00002: val_loss did not improve from 0.08859 90/90 - 1s - loss: 0.0697 - val_loss: 0.1741 - lr: 0.0010 - 742ms/epoch - 8ms/step Epoch 3/500 Epoch 00003: val_loss improved from 0.08859 to 0.03433, saving model to LSTM5.h5 90/90 - 1s - loss: 0.0543 - val_loss: 0.0343 - lr: 0.0010 - 787ms/epoch - 9ms/step Epoch 4/500 Epoch 00004: val_loss improved from 0.03433 to 0.01881, saving model to LSTM5.h5 90/90 - 1s - loss: 0.0596 - val_loss: 0.0188 - lr: 0.0010 - 739ms/epoch - 8ms/step Epoch 5/500 Epoch 00005: val_loss improved from 0.01881 to 0.01390, saving model to LSTM5.h5 90/90 - 1s - loss: 0.0722 - val_loss: 0.0139 - lr: 0.0010 - 744ms/epoch - 8ms/step Epoch 6/500 Epoch 00006: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0503 - val_loss: 0.4205 - lr: 0.0010 - 706ms/epoch - 8ms/step Epoch 7/500 Epoch 00007: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0424 - val_loss: 0.0959 - lr: 0.0010 - 699ms/epoch - 8ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0489 - val_loss: 0.5773 - lr: 0.0010 - 695ms/epoch - 8ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0383 - val_loss: 0.3410 - lr: 0.0010 - 745ms/epoch - 8ms/step Epoch 10/500 Epoch 00010: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00010: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0328 - val_loss: 0.1177 - lr: 0.0010 - 783ms/epoch - 9ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0284 - val_loss: 0.1371 - lr: 1.0000e-04 - 715ms/epoch - 8ms/step Epoch 12/500 Epoch 00012: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0272 - val_loss: 0.1484 - lr: 1.0000e-04 - 680ms/epoch - 8ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0274 - val_loss: 0.1456 - lr: 1.0000e-04 - 704ms/epoch - 8ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0284 - val_loss: 0.1484 - lr: 1.0000e-04 - 697ms/epoch - 8ms/step Epoch 15/500 Epoch 00015: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00015: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0277 - val_loss: 0.1554 - lr: 1.0000e-04 - 708ms/epoch - 8ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0245 - val_loss: 0.1553 - lr: 1.0000e-05 - 695ms/epoch - 8ms/step Epoch 17/500 Epoch 00017: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0270 - val_loss: 0.1547 - lr: 1.0000e-05 - 714ms/epoch - 8ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0238 - val_loss: 0.1542 - lr: 1.0000e-05 - 695ms/epoch - 8ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0251 - val_loss: 0.1534 - lr: 1.0000e-05 - 888ms/epoch - 10ms/step Epoch 20/500 Epoch 00020: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00020: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0260 - val_loss: 0.1523 - lr: 1.0000e-05 - 739ms/epoch - 8ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0266 - val_loss: 0.1516 - lr: 1.0000e-05 - 726ms/epoch - 8ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0263 - val_loss: 0.1502 - lr: 1.0000e-05 - 713ms/epoch - 8ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0259 - val_loss: 0.1486 - lr: 1.0000e-05 - 676ms/epoch - 8ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0236 - val_loss: 0.1482 - lr: 1.0000e-05 - 714ms/epoch - 8ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0226 - val_loss: 0.1466 - lr: 1.0000e-05 - 725ms/epoch - 8ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0235 - val_loss: 0.1464 - lr: 1.0000e-05 - 789ms/epoch - 9ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0241 - val_loss: 0.1462 - lr: 1.0000e-05 - 704ms/epoch - 8ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0256 - val_loss: 0.1468 - lr: 1.0000e-05 - 743ms/epoch - 8ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0259 - val_loss: 0.1477 - lr: 1.0000e-05 - 745ms/epoch - 8ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0242 - val_loss: 0.1481 - lr: 1.0000e-05 - 678ms/epoch - 8ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0235 - val_loss: 0.1468 - lr: 1.0000e-05 - 728ms/epoch - 8ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0263 - val_loss: 0.1465 - lr: 1.0000e-05 - 822ms/epoch - 9ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0244 - val_loss: 0.1467 - lr: 1.0000e-05 - 770ms/epoch - 9ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0253 - val_loss: 0.1454 - lr: 1.0000e-05 - 719ms/epoch - 8ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0228 - val_loss: 0.1446 - lr: 1.0000e-05 - 750ms/epoch - 8ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0229 - val_loss: 0.1412 - lr: 1.0000e-05 - 727ms/epoch - 8ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0221 - val_loss: 0.1392 - lr: 1.0000e-05 - 681ms/epoch - 8ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0255 - val_loss: 0.1384 - lr: 1.0000e-05 - 700ms/epoch - 8ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0223 - val_loss: 0.1380 - lr: 1.0000e-05 - 699ms/epoch - 8ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0238 - val_loss: 0.1364 - lr: 1.0000e-05 - 714ms/epoch - 8ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0239 - val_loss: 0.1373 - lr: 1.0000e-05 - 691ms/epoch - 8ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0241 - val_loss: 0.1377 - lr: 1.0000e-05 - 696ms/epoch - 8ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0253 - val_loss: 0.1386 - lr: 1.0000e-05 - 730ms/epoch - 8ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0230 - val_loss: 0.1363 - lr: 1.0000e-05 - 702ms/epoch - 8ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0241 - val_loss: 0.1346 - lr: 1.0000e-05 - 702ms/epoch - 8ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0227 - val_loss: 0.1344 - lr: 1.0000e-05 - 718ms/epoch - 8ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0231 - val_loss: 0.1331 - lr: 1.0000e-05 - 797ms/epoch - 9ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0231 - val_loss: 0.1303 - lr: 1.0000e-05 - 763ms/epoch - 8ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0229 - val_loss: 0.1301 - lr: 1.0000e-05 - 718ms/epoch - 8ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0219 - val_loss: 0.1300 - lr: 1.0000e-05 - 791ms/epoch - 9ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0232 - val_loss: 0.1308 - lr: 1.0000e-05 - 730ms/epoch - 8ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0210 - val_loss: 0.1310 - lr: 1.0000e-05 - 710ms/epoch - 8ms/step Epoch 53/500 Epoch 00053: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0220 - val_loss: 0.1318 - lr: 1.0000e-05 - 692ms/epoch - 8ms/step Epoch 54/500 Epoch 00054: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0233 - val_loss: 0.1301 - lr: 1.0000e-05 - 706ms/epoch - 8ms/step Epoch 55/500 Epoch 00055: val_loss did not improve from 0.01390 90/90 - 1s - loss: 0.0227 - val_loss: 0.1280 - lr: 1.0000e-05 - 708ms/epoch - 8ms/step Epoch 00055: early stopping
SMA Prediction vs Close: 46.27% Accuracy Prediction vs Prediction: 45.52% Accuracy MSE: 22.905457367129987 RMSE: 4.785964622427749 MAPE: 4.003866138329542 EMA Prediction vs Close: 54.1% Accuracy Prediction vs Prediction: 44.03% Accuracy MSE: 24.329194491765858 RMSE: 4.932463328983385 MAPE: 4.099346553838579 WMA Prediction vs Close: 54.48% Accuracy Prediction vs Prediction: 45.15% Accuracy MSE: 26.731339639305343 RMSE: 5.170235936522176 MAPE: 4.142288801040536 DEMA Prediction vs Close: 48.13% Accuracy Prediction vs Prediction: 48.51% Accuracy MSE: 20.555600166870196 RMSE: 4.53382842274277 MAPE: 3.6522177332314283 KAMA Prediction vs Close: 57.84% Accuracy Prediction vs Prediction: 44.78% Accuracy MSE: 36.658985139129086 RMSE: 6.054666393710646 MAPE: 4.91375972579294 MIDPOINT Prediction vs Close: 52.99% Accuracy Prediction vs Prediction: 44.4% Accuracy MSE: 46.211728483578064 RMSE: 6.797920894183608 MAPE: 5.510818514624332 T3 Prediction vs Close: 52.61% Accuracy Prediction vs Prediction: 45.9% Accuracy MSE: 39.422414094648474 RMSE: 6.278727107833918 MAPE: 5.177910469752962 TEMA Prediction vs Close: 47.76% Accuracy Prediction vs Prediction: 51.87% Accuracy MSE: 27.219221961342864 RMSE: 5.217204420122223 MAPE: 4.028826161355838 Runtime: mins: 54.95112048581668
Architecture Used¶
In [117]:
from google.colab import files
import cv2
uploaded = files.upload()
In [118]:
img = cv2.imread('Experiment5.png')
plt.figure(figsize=(20,10))
plt.axis("off")
plt.title('LSTM Architecture '+imgfile,fontsize=18)
plt.imshow(img)
Out[118]:
<matplotlib.image.AxesImage at 0x7f75dc11ff50>
Model Plots¶
In [108]:
with open('simulation5_data.json') as json_file:
simulation5 = json.load(json_file)
fileimg = 'Experiment5'
In [109]:
for i in range(len(list(simulation5.keys()))):
SIM = list(simulation5.keys())[i]
plot_train(simulation5,SIM)
plot_test(simulation5,SIM)
----- Train RMSE for SMA ----- 8.067881033957713 ----- Train_MSE_LSTM for SMA ----- 65.09070437809457 ----- Train MAE LSTM for SMA ----- 7.037769857791742
----- Test RMSE for SMA----- 4.785964622427749 ----- Test_MSE_LSTM for SMA----- 22.905457367129987 ----- Test_MAE_LSTM for SMA----- 4.003866138329542
----- Train RMSE for EMA ----- 9.103310494431582 ----- Train_MSE_LSTM for EMA ----- 82.8702619580282 ----- Train MAE LSTM for EMA ----- 7.942187929793381
----- Test RMSE for EMA----- 4.932463328983385 ----- Test_MSE_LSTM for EMA----- 24.329194491765858 ----- Test_MAE_LSTM for EMA----- 4.099346553838579
----- Train RMSE for WMA ----- 9.385901722462174 ----- Train_MSE_LSTM for WMA ----- 88.09515114371841 ----- Train MAE LSTM for WMA ----- 8.302396097763367
----- Test RMSE for WMA----- 5.170235936522176 ----- Test_MSE_LSTM for WMA----- 26.731339639305343 ----- Test_MAE_LSTM for WMA----- 4.142288801040536
----- Train RMSE for DEMA ----- 11.174629479833136 ----- Train_MSE_LSTM for DEMA ----- 124.8723440115558 ----- Train MAE LSTM for DEMA ----- 9.963720709179979
----- Test RMSE for DEMA----- 4.53382842274277 ----- Test_MSE_LSTM for DEMA----- 20.555600166870196 ----- Test_MAE_LSTM for DEMA----- 3.6522177332314283
----- Train RMSE for KAMA ----- 9.612463288091494 ----- Train_MSE_LSTM for KAMA ----- 92.39945046490672 ----- Train MAE LSTM for KAMA ----- 8.617087029614533
----- Test RMSE for KAMA----- 6.054666393710646 ----- Test_MSE_LSTM for KAMA----- 36.658985139129086 ----- Test_MAE_LSTM for KAMA----- 4.91375972579294
----- Train RMSE for MIDPOINT ----- 8.63124538592405 ----- Train_MSE_LSTM for MIDPOINT ----- 74.49839691203519 ----- Train MAE LSTM for MIDPOINT ----- 7.641012681592809
----- Test RMSE for MIDPOINT----- 6.797920894183608 ----- Test_MSE_LSTM for MIDPOINT----- 46.211728483578064 ----- Test_MAE_LSTM for MIDPOINT----- 5.510818514624332
----- Train RMSE for T3 ----- 10.712175422308457 ----- Train_MSE_LSTM for T3 ----- 114.75070227830938 ----- Train MAE LSTM for T3 ----- 9.554836890139239
----- Test RMSE for T3----- 6.278727107833918 ----- Test_MSE_LSTM for T3----- 39.422414094648474 ----- Test_MAE_LSTM for T3----- 5.177910469752962
----- Train RMSE for TEMA ----- 6.84563619096288 ----- Train_MSE_LSTM for TEMA ----- 46.86273485902077 ----- Train MAE LSTM for TEMA ----- 4.587888858994552
----- Test RMSE for TEMA----- 5.217204420122223 ----- Test_MSE_LSTM for TEMA----- 27.219221961342864 ----- Test_MAE_LSTM for TEMA----- 4.028826161355838
Arima w Exogenous Variable Multistep MutiVariate LSTM Hybrid Model Experiment 6¶
In [120]:
def get_arima_exog(dataframe,original_data, train_len, test_len):
# prepare train and test data for exogenous vr
X_value = pd.DataFrame(low_vol.iloc[:, :])
y_value = pd.DataFrame(low_vol.iloc[:, 3])
X_scaler = MinMaxScaler(feature_range=(-1, 1))
y_scaler = MinMaxScaler(feature_range=(-1, 1))
X_scaler.fit(X_value)
y_scaler.fit(y_value)
X_scale_dataset = X_scaler.fit_transform(X_value)
y_scale_dataset = y_scaler.fit_transform(y_value)
# Get data and check shape
# X, y, yc = get_X_y(X_scale_dataset, y_scale_dataset)#X will be of shape 224 X 3 X 21 (each 3 X 21 array will be 3 days' worth of data). yc will have the corresponding closing price value
# pdb.set_trace()
X_train, X_test, = split_train_test(X_scale_dataset)
y_train, y_test, = split_train_test(y_scale_dataset)
yc_train,yc_test = split_train_test(low_vol_data)
yc = yc_test.values.tolist()
y_train_list = y_train.flatten().tolist()
y_test_list = y_test.flatten().tolist()
# yc_train, yc_test, = split_train_test(original_data)
index_train, index_test, = predict_index(dataset_final, X_train, n_steps_in, n_steps_out)
# Initialize model
model = auto_arima(y_train_list,exogenous = X_train,trace=True, error_action='ignore', start_p=1,start_q=1,max_p=3,max_q=3,d=3,
suppress_warnings=True,stepwise=True,seasonal=True)
# Determine model parameters
print(model.summary())
model.fit(y_train_list,maxiter=200)
order = model.get_params()['order']
print('ARIMA order:', order, '\n')
# Genereate predictions
prediction = []
for i in range(len(y_test_list)):
model = pmdarima.ARIMA(order=order)
model.fit(y_train_list)
# print('working on', i+1, 'of', len(y_test), '-- ' + str(int(100 * (i + 1) / len(y_test))) + '% complete')
prediction.append(model.predict()[0])
y_train_list.append(y_test_list[i])
predictionte = y_scaler.inverse_transform(np.array(prediction).reshape(-1,1))
y_test_ = y_scaler.inverse_transform(np.array(y_test_list).reshape(-1,1))
# Generate error data
mse = mean_squared_error(yc_test, predictionte)
rmse = mse ** 0.5
mae = mean_absolute_error(y_test_ , predictionte )
return yc,predictionte.flatten().tolist(), mse, rmse, mae
In [122]:
def get_lstm(data,original_data, train_len, test_len,img_file,ma ,lstm_len=3):
# prepare train and test data
X_value = pd.DataFrame(data.iloc[:, :])
y_value = pd.DataFrame(data.iloc[:, 3])
X_scaler = MinMaxScaler(feature_range=(-1, 1))
y_scaler = MinMaxScaler(feature_range=(-1, 1))
X_scaler.fit(X_value)
y_scaler.fit(y_value)
# Get data and check shape
X, y, yc = get_X_y(X_scale_dataset, y_scale_dataset)#X will be of shape 224 X 3 X 21 (each 3 X 21 array will be 3 days' worth of data). yc will have the corresponding closing price value
# pdb.set_trace()
X_train, X_test, = split_train_test(X)
y_train, y_test, = split_train_test(y)
# yc_train, yc_test, = split_train_test(original_data)
index_train, index_test, = predict_index(dataset_final, X_train, n_steps_in, n_steps_out)
det =20
input_dim = X_train.shape[1]#3
feature_size = X_train.shape[2]#24
output_dim = y_train.shape[1]#1
# Option 1
# Set up & fit LSTM RNN
# model = Sequential()
# model.add(LSTM(256, activation='relu', kernel_initializer='he_normal', input_shape=(input_dim, feature_size)))
# model.add(Dense(units=64,activation='relu'))
# model.add(Dropout(0.5))
# model.add(Dense(units=output_dim))
# model.compile(optimizer=Adam(learning_rate = 0.001), loss='mse')
# ## Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
pyplot.show()
# # option 2
model = Sequential()
model.add(Bidirectional(LSTM(units= 128), input_shape=(input_dim, feature_size)))
model.add(Dense(64))
model.add(Dense(units=output_dim))
model.compile(optimizer=Adam(lr = 0.001), loss='mean_squared_error', metrics=['accuracy'])
# Common code
callbacks = [
EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
ModelCheckpoint('LSTM6.h5', verbose=1, save_best_only=True, save_weights_only=True)]
fname1 = img_file+'.png'
tensorflow.keras.utils.plot_model(
model, to_file=fname1, show_shapes=True, show_dtype=False,
show_layer_names=True, expand_nested=False, dpi=96,
layer_range=None, show_layer_activations=False
)
history = model.fit(X_train, y_train, epochs=500, batch_size=int( optimized_period[ma]), verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# plot loss
fname2 = img_file+'-'+ma
plt.title(img_file+'-'+ma+' Loss')
plt.xlabel("Epochs")
plt.ylabel("Loss")
pyplot.plot(history.history['loss'], label='train')
pyplot.plot(history.history['val_loss'], label='validation')
pyplot.legend()
pyplot.savefig(fname2+'.png',dpi='figure')
pyplot.show()
# Option 3
# define custom activation
#
# class Double_Tanh(Activation):
# def __init__(self, activation, **kwargs):
# super(Double_Tanh, self).__init__(activation, **kwargs)
# self.__name__ = 'double_tanh'
# def double_tanh(x):
# return (K.tanh(x) * 2)
# get_custom_objects().update({'double_tanh':Double_Tanh(double_tanh)})
# # Model Generation
# model = Sequential()
# #check https://machinelearningmastery.com/use-weight-regularization-lstm-networks-time-series-forecasting/
# model.add(LSTM(25, input_shape=(input_dim, feature_size), dropout=0.2, kernel_regularizer=l1_l2(0.00,0.00), bias_regularizer=l1_l2(0.00,0.00)))
# model.add(Dense(1))
# model.add(Activation(double_tanh))
# model.compile(loss='mean_squared_error', optimizer='adam', metrics=['mse', 'mae'])
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# Option 4
# Set up & fit LSTM RNN
# model = Sequential()
# model.add(LSTM(units=lstm_len, return_sequences=True, input_shape=(x_train.shape[1], 1)))
# model.add(LSTM(units=int(lstm_len/2)))
# model.add(Dense(1, activation='sigmoid'))
# model.compile(loss='mean_squared_error', optimizer='adam')
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# Generate predictions
predictiontr = model.predict(X_train, verbose=0)
predictiontr = y_scaler.inverse_transform(predictiontr).tolist()
outputtr = []
for i in range(len(predictiontr)):
outputtr.extend(predictiontr[i])
predictiontr = outputtr
# Generate error data
## replace with yc , xtest generated by new multistep method
mse_tr = mean_squared_error(y_train, predictiontr)
rmse_tr = mse_tr ** 0.5
# mape = mean_absolute_percentage_error(X_test, pd.Series(predictiontr))
mae_tr = mean_absolute_error(y_train, pd.Series(predictiontr))
# Original_tr = pd.Series(yc_train)
Original_tr = y_scaler.inverse_transform(y_train).flatten().tolist()
predictionte = model.predict(X_test, verbose=0)
predictionte = (y_scaler.inverse_transform(predictionte)-det).tolist()
outputte = []
for i in range(len(predictionte)):
outputte.extend(predictionte[i])
predictionte = outputte
# Generate error data
mse_te = mean_squared_error(y_test, predictionte)
rmse_te = mse_te ** 0.5
# mape = mean_absolute_percentage_error(X_test, pd.Series(predictionte))
mae_te = mean_absolute_error(y_test, pd.Series(predictionte))
# Original_te = pd.Series(yc_test)
Original_te = y_scaler.inverse_transform(y_test).flatten().tolist()
return Original_tr, predictiontr, mse_tr, rmse_tr,mae_tr,Original_te,predictionte, mse_te,rmse_te,mae_te
In [123]:
if __name__ == '__main__':
start_time = timeit.default_timer()
simulation6 = {}
imgfile = 'Experiment6'
for ma in optimized_period:
print(ma)
print(functions[ma])
print ( int( optimized_period[ma]))
# if ma == 'SMA':
low_vol = df.apply(lambda c: functions[ma](c, timeperiod = int( optimized_period[ma])))
low_vol = low_vol.fillna(0)
low_vol_data = df['close']
high_vol = pd.DataFrame()
df2 = df.copy()
for i in df2.columns:
if i in low_vol.columns:
high_vol[i] = df2[i].subtract(low_vol[i], fill_value=0)
high_vol_data = df['close']
## *****************************************************
# Generate ARIMA and LSTM predictions
print('\nWorking on ' + ma + ' predictions')
try:
print('parameters used : ', train_len, test_len)
low_vol_Original, low_vol_prediction, low_vol_mse, low_vol_rmse,low_vol_mae = get_arima_exog(low_vol,low_vol_data, train_len, test_len)
except:
print('ARIMA error, skipping to next MA type')
continue
Original_tr, predictiontr, mse_tr, rmse_tr,mae_tr,high_vol_Original, high_vol_prediction, high_vol_mse, high_vol_rmse,high_vol_mae, = get_lstm(high_vol,high_vol_data, train_len, test_len,imgfile,ma)
final_prediction_tr = df['close'].head(train_len).values + pd.Series(predictiontr) # ignoring first 3 steps
mse_ftr = mean_squared_error(df['close'].head(train_len).values,final_prediction_tr.values)
rmse_ftr = mse_ftr ** 0.5
mape_ftr = mean_absolute_percentage_error(df['close'].head(train_len).reset_index(drop=True), final_prediction_tr)
mae_ftr = mean_absolute_error(df['close'].head(train_len).reset_index(drop=True), final_prediction_tr)
final_prediction = pd.Series(low_vol_prediction[3:]) + pd.Series(high_vol_prediction)
mse = mean_squared_error(df['close'].tail(test_len).values,final_prediction.values)
rmse = mse ** 0.5
mape = mean_absolute_percentage_error(df['close'].tail(test_len).reset_index(drop=True), final_prediction)
mae = mean_absolute_error(df['close'].tail(test_len).reset_index(drop=True), final_prediction)
# Generate prediction accuracy
actual = df['close'].tail(test_len).values
result_1 = []
result_2 = []
for i in range(1, len(final_prediction)):
# Compare prediction to previous close price
if final_prediction[i] > actual[i-1] and actual[i] > actual[i-1]:
result_1.append(1)
elif final_prediction[i] < actual[i-1] and actual[i] < actual[i-1]:
result_1.append(1)
else:
result_1.append(0)
# Compare prediction to previous prediction
if final_prediction[i] > final_prediction[i-1] and actual[i] > actual[i-1]:
result_2.append(1)
elif final_prediction[i] < final_prediction[i-1] and actual[i] < actual[i-1]:
result_2.append(1)
else:
result_2.append(0)
accuracy_1 = np.mean(result_1)
accuracy_2 = np.mean(result_2)
simulation6[ma] = {'low_vol': {'original':list(low_vol_Original), 'prediction': list(low_vol_prediction) , 'mse': low_vol_mse,
'rmse': low_vol_rmse, 'mae' : low_vol_mae},
'high_vol': {'original':list(high_vol_Original),'prediction': list(high_vol_prediction), 'mse': high_vol_mse,
'rmse': high_vol_rmse, 'mae' : high_vol_mae},
'final_tr': {'original':df['close'].head(train_len).tolist(),'prediction': final_prediction_tr.values.tolist(), 'mse': mse_ftr,
'rmse': rmse_ftr, 'mae' : mae_ftr},
'final': {'original': df['close'].tail(test_len).tolist(), 'prediction': final_prediction.values.tolist(), 'mse': mse,
'rmse': rmse, 'mae': mae },
'accuracy': {'prediction vs close': accuracy_1, 'prediction vs prediction': accuracy_2}}
# save simulation data here as checkpoint
with open('simulation6_data.json', 'w') as fp:
json.dump(simulation6, fp)
for ma in simulation6.keys():
print('\n' + ma)
print('Prediction vs Close:\t\t' + str(round(100*simulation6[ma]['accuracy']['prediction vs close'], 2))
+ '% Accuracy')
print('Prediction vs Prediction:\t' + str(round(100*simulation6[ma]['accuracy']['prediction vs prediction'], 2))
+ '% Accuracy')
print('MSE:\t', simulation6[ma]['final']['mse'],
'\nRMSE:\t', simulation6[ma]['final']['rmse'],
'\nMAPE:\t', simulation6[ma]['final']['mae'])#,
# '\nMAPE:\t', simulation[ma]['final']['mape'])
# else:
# break
elapsed = timeit.default_timer() - start_time
print('Runtime: mins:',elapsed/60)
SMA
SMA([input_arrays], [timeperiod=30])
Simple Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
17
Working on SMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.786, Time=3.31 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.592, Time=4.55 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-15578.394, Time=8.68 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.592, Time=7.49 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16966.361, Time=9.62 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-16121.635, Time=10.86 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17214.069, Time=13.36 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-14562.592, Time=9.23 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-14572.319, Time=10.05 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-14403.474, Time=42.41 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 119.573 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8634.035
Date: Sun, 12 Dec 2021 AIC -17214.069
Time: 16:07:00 BIC -17087.416
Sample: 0 HQIC -17165.429
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -4.257e-09 9.55e-06 -0.000 1.000 -1.87e-05 1.87e-05
x2 -4.256e-09 9.56e-06 -0.000 1.000 -1.87e-05 1.87e-05
x3 -4.313e-09 9.62e-06 -0.000 1.000 -1.89e-05 1.88e-05
x4 1.0000 9.61e-06 1.04e+05 0.000 1.000 1.000
x5 -3.891e-09 9.14e-06 -0.000 1.000 -1.79e-05 1.79e-05
x6 -1.122e-08 1.03e-05 -0.001 0.999 -2.03e-05 2.03e-05
x7 -4.223e-09 9.54e-06 -0.000 1.000 -1.87e-05 1.87e-05
x8 -4.234e-09 9.55e-06 -0.000 1.000 -1.87e-05 1.87e-05
x9 -1.626e-10 6.54e-07 -0.000 1.000 -1.28e-06 1.28e-06
x10 -6.831e-10 2.91e-06 -0.000 1.000 -5.7e-06 5.7e-06
x11 -4.115e-09 9.41e-06 -0.000 1.000 -1.84e-05 1.84e-05
x12 -4.303e-09 9.62e-06 -0.000 1.000 -1.89e-05 1.88e-05
x13 -4.288e-09 9.6e-06 -0.000 1.000 -1.88e-05 1.88e-05
x14 -3.749e-08 2.81e-05 -0.001 0.999 -5.51e-05 5.5e-05
x15 -5.032e-09 1.04e-05 -0.000 1.000 -2.04e-05 2.03e-05
x16 -3.685e-09 9e-06 -0.000 1.000 -1.76e-05 1.76e-05
x17 -3.286e-09 8.45e-06 -0.000 1.000 -1.66e-05 1.66e-05
x18 -1.22e-08 1.59e-05 -0.001 0.999 -3.11e-05 3.11e-05
x19 -5.685e-09 1.1e-05 -0.001 1.000 -2.16e-05 2.16e-05
x20 -1.42e-08 1.69e-05 -0.001 0.999 -3.32e-05 3.32e-05
x21 -5.194e-08 3.31e-05 -0.002 0.999 -6.49e-05 6.48e-05
x22 -2.548e-08 2.31e-05 -0.001 0.999 -4.53e-05 4.52e-05
x23 -3.534e-08 2.73e-05 -0.001 0.999 -5.35e-05 5.34e-05
x24 -1.566e-08 1.8e-05 -0.001 0.999 -3.53e-05 3.53e-05
ma.L1 -1.3899 4.98e-09 -2.79e+08 0.000 -1.390 -1.390
ma.L2 0.4032 4.98e-09 8.09e+07 0.000 0.403 0.403
sigma2 7.635e-11 6.92e-11 1.103 0.270 -5.93e-11 2.12e-10
===================================================================================
Ljung-Box (L1) (Q): 68.48 Jarque-Bera (JB): 5579791.06
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 10.12
Prob(H) (two-sided): 0.00 Kurtosis: 410.36
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.69e+24. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
/usr/local/lib/python3.7/dist-packages/keras/optimizer_v2/adam.py:105: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead. super(Adam, self).__init__(name, **kwargs)
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.00838, saving model to LSTM6.h5 48/48 - 4s - loss: 0.1905 - accuracy: 0.0000e+00 - val_loss: 0.0084 - val_accuracy: 0.0037 - lr: 0.0010 - 4s/epoch - 81ms/step Epoch 2/500 Epoch 00002: val_loss did not improve from 0.00838 48/48 - 0s - loss: 0.0400 - accuracy: 0.0000e+00 - val_loss: 0.0181 - val_accuracy: 0.0037 - lr: 0.0010 - 245ms/epoch - 5ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.00838 48/48 - 0s - loss: 0.0195 - accuracy: 0.0000e+00 - val_loss: 0.0654 - val_accuracy: 0.0037 - lr: 0.0010 - 260ms/epoch - 5ms/step Epoch 4/500 Epoch 00004: val_loss improved from 0.00838 to 0.00497, saving model to LSTM6.h5 48/48 - 0s - loss: 0.0279 - accuracy: 0.0000e+00 - val_loss: 0.0050 - val_accuracy: 0.0037 - lr: 0.0010 - 314ms/epoch - 7ms/step Epoch 5/500 Epoch 00005: val_loss improved from 0.00497 to 0.00441, saving model to LSTM6.h5 48/48 - 0s - loss: 0.0026 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 0.0010 - 309ms/epoch - 6ms/step Epoch 6/500 Epoch 00006: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0025 - accuracy: 0.0000e+00 - val_loss: 0.0080 - val_accuracy: 0.0037 - lr: 0.0010 - 255ms/epoch - 5ms/step Epoch 7/500 Epoch 00007: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0018 - accuracy: 0.0000e+00 - val_loss: 0.0108 - val_accuracy: 0.0037 - lr: 0.0010 - 262ms/epoch - 5ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0018 - accuracy: 0.0000e+00 - val_loss: 0.0140 - val_accuracy: 0.0037 - lr: 0.0010 - 267ms/epoch - 6ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0020 - accuracy: 0.0000e+00 - val_loss: 0.0166 - val_accuracy: 0.0037 - lr: 0.0010 - 271ms/epoch - 6ms/step Epoch 10/500 Epoch 00010: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00010: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0022 - accuracy: 0.0000e+00 - val_loss: 0.0143 - val_accuracy: 0.0037 - lr: 0.0010 - 282ms/epoch - 6ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0046 - accuracy: 0.0000e+00 - val_loss: 0.0198 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 271ms/epoch - 6ms/step Epoch 12/500 Epoch 00012: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0051 - accuracy: 0.0000e+00 - val_loss: 0.0117 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 298ms/epoch - 6ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0028 - accuracy: 0.0000e+00 - val_loss: 0.0099 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 279ms/epoch - 6ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0021 - accuracy: 0.0000e+00 - val_loss: 0.0090 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 256ms/epoch - 5ms/step Epoch 15/500 Epoch 00015: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00015: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0018 - accuracy: 0.0000e+00 - val_loss: 0.0086 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 269ms/epoch - 6ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0082 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 294ms/epoch - 6ms/step Epoch 17/500 Epoch 00017: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0080 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 251ms/epoch - 5ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0078 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 254ms/epoch - 5ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0077 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 258ms/epoch - 5ms/step Epoch 20/500 Epoch 00020: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00020: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0076 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 265ms/epoch - 6ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0075 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 278ms/epoch - 6ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0075 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 255ms/epoch - 5ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0074 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 301ms/epoch - 6ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0074 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 250ms/epoch - 5ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0073 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 272ms/epoch - 6ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0073 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 270ms/epoch - 6ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0073 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 250ms/epoch - 5ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0072 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 288ms/epoch - 6ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0072 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 272ms/epoch - 6ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0071 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 258ms/epoch - 5ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0071 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 272ms/epoch - 6ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0070 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 253ms/epoch - 5ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0070 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 263ms/epoch - 5ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0070 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 253ms/epoch - 5ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0069 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 283ms/epoch - 6ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0069 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 253ms/epoch - 5ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0068 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 276ms/epoch - 6ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0068 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 251ms/epoch - 5ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0067 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 292ms/epoch - 6ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0067 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 252ms/epoch - 5ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0066 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 276ms/epoch - 6ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0066 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 294ms/epoch - 6ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0065 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 266ms/epoch - 6ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 262ms/epoch - 5ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 269ms/epoch - 6ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0063 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 279ms/epoch - 6ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0063 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 268ms/epoch - 6ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0062 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 249ms/epoch - 5ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.00441 48/48 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0062 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 248ms/epoch - 5ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.00441 48/48 - 0s - loss: 9.9537e-04 - accuracy: 0.0000e+00 - val_loss: 0.0061 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 261ms/epoch - 5ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.00441 48/48 - 0s - loss: 9.8973e-04 - accuracy: 0.0000e+00 - val_loss: 0.0061 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 242ms/epoch - 5ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.00441 48/48 - 0s - loss: 9.8420e-04 - accuracy: 0.0000e+00 - val_loss: 0.0060 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 254ms/epoch - 5ms/step Epoch 53/500 Epoch 00053: val_loss did not improve from 0.00441 48/48 - 0s - loss: 9.7879e-04 - accuracy: 0.0000e+00 - val_loss: 0.0060 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 249ms/epoch - 5ms/step Epoch 54/500 Epoch 00054: val_loss did not improve from 0.00441 48/48 - 0s - loss: 9.7348e-04 - accuracy: 0.0000e+00 - val_loss: 0.0059 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 297ms/epoch - 6ms/step Epoch 55/500 Epoch 00055: val_loss did not improve from 0.00441 48/48 - 0s - loss: 9.6828e-04 - accuracy: 0.0000e+00 - val_loss: 0.0059 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 249ms/epoch - 5ms/step Epoch 00055: early stopping
SMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 109.05489356902385
RMSE: 10.442935103170173
MAPE: 8.732625329283675
EMA
EMA([input_arrays], [timeperiod=30])
Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
51
Working on EMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.778, Time=3.00 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.589, Time=4.60 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-14606.447, Time=6.14 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.589, Time=6.87 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-15343.613, Time=9.59 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-15047.583, Time=13.16 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-16858.964, Time=11.54 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=-17024.022, Time=6.08 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=-16998.618, Time=3.44 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=-17081.451, Time=7.17 sec
ARIMA(3,3,2)(0,0,0)[0] : AIC=inf, Time=17.36 sec
/usr/local/lib/python3.7/dist-packages/statsmodels/tsa/statespace/sarimax.py:1906: RuntimeWarning: divide by zero encountered in reciprocal return np.roots(self.polynomial_reduced_ma)**-1
ARIMA(2,3,2)(0,0,0)[0] : AIC=-16997.990, Time=3.81 sec
ARIMA(3,3,1)(0,0,0)[0] intercept : AIC=-16992.667, Time=4.71 sec
Best model: ARIMA(3,3,1)(0,0,0)[0]
Total fit time: 97.502 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 1) Log Likelihood 8569.726
Date: Sun, 12 Dec 2021 AIC -17081.451
Time: 16:12:48 BIC -16945.417
Sample: 0 HQIC -17029.208
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -2.316e-10 9.89e-05 -2.34e-06 1.000 -0.000 0.000
x2 -2.309e-10 9.88e-05 -2.34e-06 1.000 -0.000 0.000
x3 -2.325e-10 9.91e-05 -2.35e-06 1.000 -0.000 0.000
x4 1.0000 9.9e-05 1.01e+04 0.000 1.000 1.000
x5 -2.108e-10 9.43e-05 -2.24e-06 1.000 -0.000 0.000
x6 -7.997e-10 0.000 -4.63e-06 1.000 -0.000 0.000
x7 -2.295e-10 9.85e-05 -2.33e-06 1.000 -0.000 0.000
x8 -2.244e-10 9.74e-05 -2.3e-06 1.000 -0.000 0.000
x9 -1.166e-11 1.98e-05 -5.9e-07 1.000 -3.87e-05 3.87e-05
x10 -4.454e-11 4.19e-05 -1.06e-06 1.000 -8.22e-05 8.22e-05
x11 -2.219e-10 9.68e-05 -2.29e-06 1.000 -0.000 0.000
x12 -2.264e-10 9.8e-05 -2.31e-06 1.000 -0.000 0.000
x13 -2.315e-10 9.89e-05 -2.34e-06 1.000 -0.000 0.000
x14 -1.767e-09 0.000 -6.47e-06 1.000 -0.001 0.001
x15 -2.096e-10 9.38e-05 -2.23e-06 1.000 -0.000 0.000
x16 -5.257e-10 0.000 -3.5e-06 1.000 -0.000 0.000
x17 -2.143e-10 9.53e-05 -2.25e-06 1.000 -0.000 0.000
x18 -3.776e-11 3.61e-05 -1.05e-06 1.000 -7.08e-05 7.08e-05
x19 -2.52e-10 0.000 -2.41e-06 1.000 -0.000 0.000
x20 -2.417e-10 9.51e-05 -2.54e-06 1.000 -0.000 0.000
x21 -3.16e-09 0.000 -8.64e-06 1.000 -0.001 0.001
x22 -2.955e-09 0.000 -8.32e-06 1.000 -0.001 0.001
x23 -1.664e-09 0.000 -6.29e-06 1.000 -0.001 0.001
x24 -1.568e-09 0.000 -6.07e-06 1.000 -0.001 0.001
ar.L1 -0.4923 1.2e-09 -4.09e+08 0.000 -0.492 -0.492
ar.L2 -0.1923 7e-10 -2.75e+08 0.000 -0.192 -0.192
ar.L3 -0.0461 3.32e-10 -1.39e+08 0.000 -0.046 -0.046
ma.L1 -0.7077 2.73e-09 -2.59e+08 0.000 -0.708 -0.708
sigma2 8.99e-11 6.96e-11 1.291 0.197 -4.66e-11 2.26e-10
===================================================================================
Ljung-Box (L1) (Q): 55.51 Jarque-Bera (JB): 4268313.90
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 5.44
Prob(H) (two-sided): 0.00 Kurtosis: 359.56
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.36e+28. Standard errors may be unstable.
ARIMA order: (3, 3, 1)
/usr/local/lib/python3.7/dist-packages/keras/optimizer_v2/adam.py:105: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead. super(Adam, self).__init__(name, **kwargs)
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.01247, saving model to LSTM6.h5 16/16 - 4s - loss: 0.1231 - accuracy: 0.0000e+00 - val_loss: 0.0125 - val_accuracy: 0.0037 - lr: 0.0010 - 4s/epoch - 225ms/step Epoch 2/500 Epoch 00002: val_loss did not improve from 0.01247 16/16 - 0s - loss: 0.0716 - accuracy: 0.0000e+00 - val_loss: 0.0196 - val_accuracy: 0.0037 - lr: 0.0010 - 91ms/epoch - 6ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.01247 16/16 - 0s - loss: 0.0081 - accuracy: 0.0000e+00 - val_loss: 0.0408 - val_accuracy: 0.0037 - lr: 0.0010 - 138ms/epoch - 9ms/step Epoch 4/500 Epoch 00004: val_loss improved from 0.01247 to 0.00713, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0251 - accuracy: 0.0000e+00 - val_loss: 0.0071 - val_accuracy: 0.0037 - lr: 0.0010 - 130ms/epoch - 8ms/step Epoch 5/500 Epoch 00005: val_loss did not improve from 0.00713 16/16 - 0s - loss: 0.0132 - accuracy: 0.0000e+00 - val_loss: 0.0361 - val_accuracy: 0.0037 - lr: 0.0010 - 104ms/epoch - 6ms/step Epoch 6/500 Epoch 00006: val_loss did not improve from 0.00713 16/16 - 0s - loss: 0.0307 - accuracy: 0.0000e+00 - val_loss: 0.0107 - val_accuracy: 0.0037 - lr: 0.0010 - 106ms/epoch - 7ms/step Epoch 7/500 Epoch 00007: val_loss did not improve from 0.00713 16/16 - 0s - loss: 0.0109 - accuracy: 0.0000e+00 - val_loss: 0.0363 - val_accuracy: 0.0037 - lr: 0.0010 - 102ms/epoch - 6ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.00713 16/16 - 0s - loss: 0.0310 - accuracy: 0.0000e+00 - val_loss: 0.0097 - val_accuracy: 0.0037 - lr: 0.0010 - 96ms/epoch - 6ms/step Epoch 9/500 Epoch 00009: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00009: val_loss did not improve from 0.00713 16/16 - 0s - loss: 0.0140 - accuracy: 0.0000e+00 - val_loss: 0.0284 - val_accuracy: 0.0037 - lr: 0.0010 - 116ms/epoch - 7ms/step Epoch 10/500 Epoch 00010: val_loss did not improve from 0.00713 16/16 - 0s - loss: 0.0268 - accuracy: 0.0000e+00 - val_loss: 0.0118 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 114ms/epoch - 7ms/step Epoch 11/500 Epoch 00011: val_loss improved from 0.00713 to 0.00653, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0044 - accuracy: 0.0000e+00 - val_loss: 0.0065 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 121ms/epoch - 8ms/step Epoch 12/500 Epoch 00012: val_loss improved from 0.00653 to 0.00623, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0025 - accuracy: 0.0000e+00 - val_loss: 0.0062 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 118ms/epoch - 7ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.00623 16/16 - 0s - loss: 0.0019 - accuracy: 0.0000e+00 - val_loss: 0.0065 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 115ms/epoch - 7ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.00623 16/16 - 0s - loss: 0.0017 - accuracy: 0.0000e+00 - val_loss: 0.0065 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 114ms/epoch - 7ms/step Epoch 15/500 Epoch 00015: val_loss did not improve from 0.00623 16/16 - 0s - loss: 0.0016 - accuracy: 0.0000e+00 - val_loss: 0.0063 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 102ms/epoch - 6ms/step Epoch 16/500 Epoch 00016: val_loss improved from 0.00623 to 0.00609, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0015 - accuracy: 0.0000e+00 - val_loss: 0.0061 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 121ms/epoch - 8ms/step Epoch 17/500 Epoch 00017: val_loss improved from 0.00609 to 0.00596, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0060 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 131ms/epoch - 8ms/step Epoch 18/500 Epoch 00018: val_loss improved from 0.00596 to 0.00585, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0058 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 128ms/epoch - 8ms/step Epoch 19/500 Epoch 00019: val_loss improved from 0.00585 to 0.00574, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0057 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 126ms/epoch - 8ms/step Epoch 20/500 Epoch 00020: val_loss improved from 0.00574 to 0.00564, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0056 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 130ms/epoch - 8ms/step Epoch 21/500 Epoch 00021: val_loss improved from 0.00564 to 0.00555, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0056 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 136ms/epoch - 8ms/step Epoch 22/500 Epoch 00022: val_loss improved from 0.00555 to 0.00547, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0055 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 134ms/epoch - 8ms/step Epoch 23/500 Epoch 00023: val_loss improved from 0.00547 to 0.00540, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0054 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 142ms/epoch - 9ms/step Epoch 24/500 Epoch 00024: val_loss improved from 0.00540 to 0.00532, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0053 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 157ms/epoch - 10ms/step Epoch 25/500 Epoch 00025: val_loss improved from 0.00532 to 0.00526, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0053 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 138ms/epoch - 9ms/step Epoch 26/500 Epoch 00026: val_loss improved from 0.00526 to 0.00520, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0052 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 135ms/epoch - 8ms/step Epoch 27/500 Epoch 00027: val_loss improved from 0.00520 to 0.00514, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0051 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 142ms/epoch - 9ms/step Epoch 28/500 Epoch 00028: val_loss improved from 0.00514 to 0.00508, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0051 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 124ms/epoch - 8ms/step Epoch 29/500 Epoch 00029: val_loss improved from 0.00508 to 0.00503, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0050 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 126ms/epoch - 8ms/step Epoch 30/500 Epoch 00030: val_loss improved from 0.00503 to 0.00498, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0050 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 109ms/epoch - 7ms/step Epoch 31/500 Epoch 00031: val_loss improved from 0.00498 to 0.00494, saving model to LSTM6.h5 16/16 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0049 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 125ms/epoch - 8ms/step Epoch 32/500 Epoch 00032: val_loss improved from 0.00494 to 0.00489, saving model to LSTM6.h5 16/16 - 0s - loss: 9.9457e-04 - accuracy: 0.0000e+00 - val_loss: 0.0049 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 134ms/epoch - 8ms/step Epoch 33/500 Epoch 00033: val_loss improved from 0.00489 to 0.00485, saving model to LSTM6.h5 16/16 - 0s - loss: 9.8588e-04 - accuracy: 0.0000e+00 - val_loss: 0.0049 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 126ms/epoch - 8ms/step Epoch 34/500 Epoch 00034: val_loss improved from 0.00485 to 0.00482, saving model to LSTM6.h5 16/16 - 0s - loss: 9.7792e-04 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 137ms/epoch - 9ms/step Epoch 35/500 Epoch 00035: val_loss improved from 0.00482 to 0.00478, saving model to LSTM6.h5 16/16 - 0s - loss: 9.7058e-04 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 140ms/epoch - 9ms/step Epoch 36/500 Epoch 00036: val_loss improved from 0.00478 to 0.00475, saving model to LSTM6.h5 16/16 - 0s - loss: 9.6376e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 118ms/epoch - 7ms/step Epoch 37/500 Epoch 00037: val_loss improved from 0.00475 to 0.00472, saving model to LSTM6.h5 16/16 - 0s - loss: 9.5736e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 120ms/epoch - 8ms/step Epoch 38/500 Epoch 00038: val_loss improved from 0.00472 to 0.00469, saving model to LSTM6.h5 16/16 - 0s - loss: 9.5131e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 120ms/epoch - 8ms/step Epoch 39/500 Epoch 00039: val_loss improved from 0.00469 to 0.00466, saving model to LSTM6.h5 16/16 - 0s - loss: 9.4556e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 135ms/epoch - 8ms/step Epoch 40/500 Epoch 00040: val_loss improved from 0.00466 to 0.00464, saving model to LSTM6.h5 16/16 - 0s - loss: 9.4006e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 139ms/epoch - 9ms/step Epoch 41/500 Epoch 00041: val_loss improved from 0.00464 to 0.00461, saving model to LSTM6.h5 16/16 - 0s - loss: 9.3477e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 123ms/epoch - 8ms/step Epoch 42/500 Epoch 00042: val_loss improved from 0.00461 to 0.00460, saving model to LSTM6.h5 16/16 - 0s - loss: 9.2965e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 136ms/epoch - 9ms/step Epoch 43/500 Epoch 00043: val_loss improved from 0.00460 to 0.00458, saving model to LSTM6.h5 16/16 - 0s - loss: 9.2468e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 118ms/epoch - 7ms/step Epoch 44/500 Epoch 00044: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00044: val_loss improved from 0.00458 to 0.00456, saving model to LSTM6.h5 16/16 - 0s - loss: 9.1985e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 129ms/epoch - 8ms/step Epoch 45/500 Epoch 00045: val_loss improved from 0.00456 to 0.00456, saving model to LSTM6.h5 16/16 - 0s - loss: 9.1033e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 119ms/epoch - 7ms/step Epoch 46/500 Epoch 00046: val_loss improved from 0.00456 to 0.00456, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0981e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 131ms/epoch - 8ms/step Epoch 47/500 Epoch 00047: val_loss improved from 0.00456 to 0.00456, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0933e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 140ms/epoch - 9ms/step Epoch 48/500 Epoch 00048: val_loss improved from 0.00456 to 0.00456, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0887e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 125ms/epoch - 8ms/step Epoch 49/500 Epoch 00049: val_loss improved from 0.00456 to 0.00456, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0841e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 128ms/epoch - 8ms/step Epoch 50/500 Epoch 00050: val_loss improved from 0.00456 to 0.00456, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0795e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 132ms/epoch - 8ms/step Epoch 51/500 Epoch 00051: val_loss improved from 0.00456 to 0.00456, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0749e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 123ms/epoch - 8ms/step Epoch 52/500 Epoch 00052: val_loss improved from 0.00456 to 0.00455, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0702e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 124ms/epoch - 8ms/step Epoch 53/500 Epoch 00053: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00053: val_loss improved from 0.00455 to 0.00455, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0654e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 118ms/epoch - 7ms/step Epoch 54/500 Epoch 00054: val_loss improved from 0.00455 to 0.00455, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0606e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 137ms/epoch - 9ms/step Epoch 55/500 Epoch 00055: val_loss improved from 0.00455 to 0.00455, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0557e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 128ms/epoch - 8ms/step Epoch 56/500 Epoch 00056: val_loss improved from 0.00455 to 0.00455, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0508e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 133ms/epoch - 8ms/step Epoch 57/500 Epoch 00057: val_loss improved from 0.00455 to 0.00455, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0458e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 126ms/epoch - 8ms/step Epoch 58/500 Epoch 00058: val_loss improved from 0.00455 to 0.00455, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0408e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 179ms/epoch - 11ms/step Epoch 59/500 Epoch 00059: val_loss improved from 0.00455 to 0.00455, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0357e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 135ms/epoch - 8ms/step Epoch 60/500 Epoch 00060: val_loss improved from 0.00455 to 0.00455, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0305e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 135ms/epoch - 8ms/step Epoch 61/500 Epoch 00061: val_loss improved from 0.00455 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0254e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 132ms/epoch - 8ms/step Epoch 62/500 Epoch 00062: val_loss improved from 0.00454 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0201e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 141ms/epoch - 9ms/step Epoch 63/500 Epoch 00063: val_loss improved from 0.00454 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0148e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 140ms/epoch - 9ms/step Epoch 64/500 Epoch 00064: val_loss improved from 0.00454 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0095e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 127ms/epoch - 8ms/step Epoch 65/500 Epoch 00065: val_loss improved from 0.00454 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 9.0042e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 125ms/epoch - 8ms/step Epoch 66/500 Epoch 00066: val_loss improved from 0.00454 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 8.9987e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 128ms/epoch - 8ms/step Epoch 67/500 Epoch 00067: val_loss improved from 0.00454 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 8.9933e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 121ms/epoch - 8ms/step Epoch 68/500 Epoch 00068: val_loss improved from 0.00454 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 8.9878e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 126ms/epoch - 8ms/step Epoch 69/500 Epoch 00069: val_loss improved from 0.00454 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 8.9822e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 137ms/epoch - 9ms/step Epoch 70/500 Epoch 00070: val_loss improved from 0.00454 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 8.9767e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 124ms/epoch - 8ms/step Epoch 71/500 Epoch 00071: val_loss improved from 0.00454 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 8.9710e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 132ms/epoch - 8ms/step Epoch 72/500 Epoch 00072: val_loss improved from 0.00454 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 8.9653e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 121ms/epoch - 8ms/step Epoch 73/500 Epoch 00073: val_loss improved from 0.00454 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 8.9596e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 136ms/epoch - 9ms/step Epoch 74/500 Epoch 00074: val_loss improved from 0.00454 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 8.9539e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 123ms/epoch - 8ms/step Epoch 75/500 Epoch 00075: val_loss improved from 0.00454 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 8.9481e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 131ms/epoch - 8ms/step Epoch 76/500 Epoch 00076: val_loss improved from 0.00454 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 8.9422e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 192ms/epoch - 12ms/step Epoch 77/500 Epoch 00077: val_loss improved from 0.00454 to 0.00454, saving model to LSTM6.h5 16/16 - 0s - loss: 8.9363e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 125ms/epoch - 8ms/step Epoch 78/500 Epoch 00078: val_loss improved from 0.00454 to 0.00453, saving model to LSTM6.h5 16/16 - 0s - loss: 8.9304e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 137ms/epoch - 9ms/step Epoch 79/500 Epoch 00079: val_loss improved from 0.00453 to 0.00453, saving model to LSTM6.h5 16/16 - 0s - loss: 8.9244e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 120ms/epoch - 7ms/step Epoch 80/500 Epoch 00080: val_loss improved from 0.00453 to 0.00453, saving model to LSTM6.h5 16/16 - 0s - loss: 8.9184e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 123ms/epoch - 8ms/step Epoch 81/500 Epoch 00081: val_loss improved from 0.00453 to 0.00453, saving model to LSTM6.h5 16/16 - 0s - loss: 8.9124e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 121ms/epoch - 8ms/step Epoch 82/500 Epoch 00082: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.9063e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 101ms/epoch - 6ms/step Epoch 83/500 Epoch 00083: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.9002e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 98ms/epoch - 6ms/step Epoch 84/500 Epoch 00084: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.8940e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 107ms/epoch - 7ms/step Epoch 85/500 Epoch 00085: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.8878e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 123ms/epoch - 8ms/step Epoch 86/500 Epoch 00086: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.8815e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 108ms/epoch - 7ms/step Epoch 87/500 Epoch 00087: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.8753e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 114ms/epoch - 7ms/step Epoch 88/500 Epoch 00088: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.8689e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 100ms/epoch - 6ms/step Epoch 89/500 Epoch 00089: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.8626e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 97ms/epoch - 6ms/step Epoch 90/500 Epoch 00090: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.8562e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step Epoch 91/500 Epoch 00091: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.8497e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 146ms/epoch - 9ms/step Epoch 92/500 Epoch 00092: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.8433e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 116ms/epoch - 7ms/step Epoch 93/500 Epoch 00093: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.8368e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 114ms/epoch - 7ms/step Epoch 94/500 Epoch 00094: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.8302e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 112ms/epoch - 7ms/step Epoch 95/500 Epoch 00095: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.8236e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 110ms/epoch - 7ms/step Epoch 96/500 Epoch 00096: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.8170e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step Epoch 97/500 Epoch 00097: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.8103e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step Epoch 98/500 Epoch 00098: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.8036e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 101ms/epoch - 6ms/step Epoch 99/500 Epoch 00099: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.7969e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 100ms/epoch - 6ms/step Epoch 100/500 Epoch 00100: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.7902e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 117ms/epoch - 7ms/step Epoch 101/500 Epoch 00101: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.7834e-04 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 119ms/epoch - 7ms/step Epoch 102/500 Epoch 00102: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.7765e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 118ms/epoch - 7ms/step Epoch 103/500 Epoch 00103: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.7697e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 107ms/epoch - 7ms/step Epoch 104/500 Epoch 00104: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.7628e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 141ms/epoch - 9ms/step Epoch 105/500 Epoch 00105: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.7558e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 109ms/epoch - 7ms/step Epoch 106/500 Epoch 00106: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.7488e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 119ms/epoch - 7ms/step Epoch 107/500 Epoch 00107: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.7418e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 105ms/epoch - 7ms/step Epoch 108/500 Epoch 00108: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.7348e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 113ms/epoch - 7ms/step Epoch 109/500 Epoch 00109: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.7277e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 121ms/epoch - 8ms/step Epoch 110/500 Epoch 00110: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.7206e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 112ms/epoch - 7ms/step Epoch 111/500 Epoch 00111: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.7135e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 113ms/epoch - 7ms/step Epoch 112/500 Epoch 00112: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.7063e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 107ms/epoch - 7ms/step Epoch 113/500 Epoch 00113: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.6991e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 104ms/epoch - 7ms/step Epoch 114/500 Epoch 00114: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.6919e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step Epoch 115/500 Epoch 00115: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.6847e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 101ms/epoch - 6ms/step Epoch 116/500 Epoch 00116: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.6774e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 108ms/epoch - 7ms/step Epoch 117/500 Epoch 00117: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.6700e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 105ms/epoch - 7ms/step Epoch 118/500 Epoch 00118: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.6627e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 114ms/epoch - 7ms/step Epoch 119/500 Epoch 00119: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.6553e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 105ms/epoch - 7ms/step Epoch 120/500 Epoch 00120: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.6479e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 118ms/epoch - 7ms/step Epoch 121/500 Epoch 00121: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.6405e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step Epoch 122/500 Epoch 00122: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.6330e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 116ms/epoch - 7ms/step Epoch 123/500 Epoch 00123: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.6255e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step Epoch 124/500 Epoch 00124: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.6180e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 108ms/epoch - 7ms/step Epoch 125/500 Epoch 00125: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.6104e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 97ms/epoch - 6ms/step Epoch 126/500 Epoch 00126: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.6028e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 149ms/epoch - 9ms/step Epoch 127/500 Epoch 00127: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.5952e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 132ms/epoch - 8ms/step Epoch 128/500 Epoch 00128: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.5876e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 118ms/epoch - 7ms/step Epoch 129/500 Epoch 00129: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.5799e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 110ms/epoch - 7ms/step Epoch 130/500 Epoch 00130: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.5722e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 120ms/epoch - 7ms/step Epoch 131/500 Epoch 00131: val_loss did not improve from 0.00453 16/16 - 0s - loss: 8.5645e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 108ms/epoch - 7ms/step Epoch 00131: early stopping
SMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 109.05489356902385
RMSE: 10.442935103170173
MAPE: 8.732625329283675
EMA
Prediction vs Close: 54.85% Accuracy
Prediction vs Prediction: 44.4% Accuracy
MSE: 73.04142380966745
RMSE: 8.546427546622475
MAPE: 7.099244401385842
WMA
WMA([input_arrays], [timeperiod=30])
Weighted Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
49
Working on WMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.780, Time=3.11 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.589, Time=4.60 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-16789.784, Time=12.13 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.589, Time=7.20 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16919.987, Time=8.97 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-14616.097, Time=11.93 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17225.955, Time=17.03 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-14562.589, Time=8.91 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-15582.364, Time=18.42 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-12043.670, Time=35.94 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 128.246 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8639.977
Date: Sun, 12 Dec 2021 AIC -17225.955
Time: 16:24:12 BIC -17099.302
Sample: 0 HQIC -17177.315
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -4.802e-09 4.51e-06 -0.001 0.999 -8.84e-06 8.83e-06
x2 -4.783e-09 4.5e-06 -0.001 0.999 -8.83e-06 8.82e-06
x3 -4.811e-09 4.51e-06 -0.001 0.999 -8.85e-06 8.84e-06
x4 1.0000 4.51e-06 2.22e+05 0.000 1.000 1.000
x5 -4.353e-09 4.3e-06 -0.001 0.999 -8.43e-06 8.42e-06
x6 -1.569e-08 7.54e-06 -0.002 0.998 -1.48e-05 1.48e-05
x7 -4.75e-09 4.49e-06 -0.001 0.999 -8.8e-06 8.79e-06
x8 -4.628e-09 4.43e-06 -0.001 0.999 -8.69e-06 8.69e-06
x9 -4.733e-10 1.16e-06 -0.000 1.000 -2.27e-06 2.27e-06
x10 -7.88e-10 1.8e-06 -0.000 1.000 -3.52e-06 3.52e-06
x11 -4.609e-09 4.42e-06 -0.001 0.999 -8.68e-06 8.67e-06
x12 -4.607e-09 4.42e-06 -0.001 0.999 -8.68e-06 8.67e-06
x13 -4.792e-09 4.51e-06 -0.001 0.999 -8.84e-06 8.83e-06
x14 -3.777e-08 1.24e-05 -0.003 0.998 -2.44e-05 2.44e-05
x15 -3.99e-09 4.12e-06 -0.001 0.999 -8.08e-06 8.07e-06
x16 -1.309e-08 7.41e-06 -0.002 0.999 -1.45e-05 1.45e-05
x17 -4.789e-09 4.51e-06 -0.001 0.999 -8.85e-06 8.84e-06
x18 -2.665e-10 9.77e-07 -0.000 1.000 -1.92e-06 1.92e-06
x19 -4.919e-09 4.56e-06 -0.001 0.999 -8.94e-06 8.93e-06
x20 -4e-10 9.58e-07 -0.000 1.000 -1.88e-06 1.88e-06
x21 -6.782e-08 1.67e-05 -0.004 0.997 -3.27e-05 3.26e-05
x22 -6.03e-08 1.58e-05 -0.004 0.997 -3.09e-05 3.08e-05
x23 -3.157e-08 1.14e-05 -0.003 0.998 -2.23e-05 2.23e-05
x24 -3.671e-08 1.23e-05 -0.003 0.998 -2.41e-05 2.41e-05
ma.L1 -1.3901 5.58e-10 -2.49e+09 0.000 -1.390 -1.390
ma.L2 0.4033 5.75e-10 7.02e+08 0.000 0.403 0.403
sigma2 7.525e-11 6.92e-11 1.088 0.277 -6.03e-11 2.11e-10
===================================================================================
Ljung-Box (L1) (Q): 69.18 Jarque-Bera (JB): 6366427.21
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 12.29
Prob(H) (two-sided): 0.00 Kurtosis: 437.97
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 2.29e+25. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
/usr/local/lib/python3.7/dist-packages/keras/optimizer_v2/adam.py:105: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead. super(Adam, self).__init__(name, **kwargs)
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.04829, saving model to LSTM6.h5 17/17 - 4s - loss: 0.1049 - accuracy: 0.0000e+00 - val_loss: 0.0483 - val_accuracy: 0.0037 - lr: 0.0010 - 4s/epoch - 229ms/step Epoch 2/500 Epoch 00002: val_loss did not improve from 0.04829 17/17 - 0s - loss: 0.0452 - accuracy: 0.0000e+00 - val_loss: 0.1689 - val_accuracy: 0.0000e+00 - lr: 0.0010 - 100ms/epoch - 6ms/step Epoch 3/500 Epoch 00003: val_loss improved from 0.04829 to 0.03648, saving model to LSTM6.h5 17/17 - 0s - loss: 0.0148 - accuracy: 0.0000e+00 - val_loss: 0.0365 - val_accuracy: 0.0037 - lr: 0.0010 - 146ms/epoch - 9ms/step Epoch 4/500 Epoch 00004: val_loss improved from 0.03648 to 0.00656, saving model to LSTM6.h5 17/17 - 0s - loss: 0.0075 - accuracy: 0.0000e+00 - val_loss: 0.0066 - val_accuracy: 0.0037 - lr: 0.0010 - 147ms/epoch - 9ms/step Epoch 5/500 Epoch 00005: val_loss did not improve from 0.00656 17/17 - 0s - loss: 0.0322 - accuracy: 0.0000e+00 - val_loss: 0.0123 - val_accuracy: 0.0037 - lr: 0.0010 - 107ms/epoch - 6ms/step Epoch 6/500 Epoch 00006: val_loss did not improve from 0.00656 17/17 - 0s - loss: 0.0128 - accuracy: 0.0000e+00 - val_loss: 0.0463 - val_accuracy: 0.0037 - lr: 0.0010 - 127ms/epoch - 7ms/step Epoch 7/500 Epoch 00007: val_loss did not improve from 0.00656 17/17 - 0s - loss: 0.0221 - accuracy: 0.0000e+00 - val_loss: 0.0596 - val_accuracy: 0.0037 - lr: 0.0010 - 130ms/epoch - 8ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.00656 17/17 - 0s - loss: 0.0313 - accuracy: 0.0000e+00 - val_loss: 0.0079 - val_accuracy: 0.0037 - lr: 0.0010 - 104ms/epoch - 6ms/step Epoch 9/500 Epoch 00009: val_loss improved from 0.00656 to 0.00571, saving model to LSTM6.h5 17/17 - 0s - loss: 0.0277 - accuracy: 0.0000e+00 - val_loss: 0.0057 - val_accuracy: 0.0037 - lr: 0.0010 - 138ms/epoch - 8ms/step Epoch 10/500 Epoch 00010: val_loss did not improve from 0.00571 17/17 - 0s - loss: 0.0312 - accuracy: 0.0000e+00 - val_loss: 0.0744 - val_accuracy: 0.0000e+00 - lr: 0.0010 - 106ms/epoch - 6ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.00571 17/17 - 0s - loss: 0.0280 - accuracy: 0.0000e+00 - val_loss: 0.0155 - val_accuracy: 0.0037 - lr: 0.0010 - 115ms/epoch - 7ms/step Epoch 12/500 Epoch 00012: val_loss did not improve from 0.00571 17/17 - 0s - loss: 0.0155 - accuracy: 0.0000e+00 - val_loss: 0.0120 - val_accuracy: 0.0037 - lr: 0.0010 - 119ms/epoch - 7ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.00571 17/17 - 0s - loss: 0.0167 - accuracy: 0.0000e+00 - val_loss: 0.0214 - val_accuracy: 0.0037 - lr: 0.0010 - 114ms/epoch - 7ms/step Epoch 14/500 Epoch 00014: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00014: val_loss did not improve from 0.00571 17/17 - 0s - loss: 0.0036 - accuracy: 0.0000e+00 - val_loss: 0.0116 - val_accuracy: 0.0037 - lr: 0.0010 - 105ms/epoch - 6ms/step Epoch 15/500 Epoch 00015: val_loss did not improve from 0.00571 17/17 - 0s - loss: 0.0044 - accuracy: 0.0000e+00 - val_loss: 0.0086 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 106ms/epoch - 6ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.00571 17/17 - 0s - loss: 0.0021 - accuracy: 0.0000e+00 - val_loss: 0.0066 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 109ms/epoch - 6ms/step Epoch 17/500 Epoch 00017: val_loss did not improve from 0.00571 17/17 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0058 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 115ms/epoch - 7ms/step Epoch 18/500 Epoch 00018: val_loss improved from 0.00571 to 0.00554, saving model to LSTM6.h5 17/17 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0055 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 141ms/epoch - 8ms/step Epoch 19/500 Epoch 00019: val_loss improved from 0.00554 to 0.00540, saving model to LSTM6.h5 17/17 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0054 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 134ms/epoch - 8ms/step Epoch 20/500 Epoch 00020: val_loss improved from 0.00540 to 0.00525, saving model to LSTM6.h5 17/17 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0052 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 140ms/epoch - 8ms/step Epoch 21/500 Epoch 00021: val_loss improved from 0.00525 to 0.00510, saving model to LSTM6.h5 17/17 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0051 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 139ms/epoch - 8ms/step Epoch 22/500 Epoch 00022: val_loss improved from 0.00510 to 0.00498, saving model to LSTM6.h5 17/17 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0050 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 127ms/epoch - 7ms/step Epoch 23/500 Epoch 00023: val_loss improved from 0.00498 to 0.00489, saving model to LSTM6.h5 17/17 - 0s - loss: 9.9787e-04 - accuracy: 0.0000e+00 - val_loss: 0.0049 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 127ms/epoch - 7ms/step Epoch 24/500 Epoch 00024: val_loss improved from 0.00489 to 0.00482, saving model to LSTM6.h5 17/17 - 0s - loss: 9.8252e-04 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 122ms/epoch - 7ms/step Epoch 25/500 Epoch 00025: val_loss improved from 0.00482 to 0.00477, saving model to LSTM6.h5 17/17 - 0s - loss: 9.7008e-04 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 145ms/epoch - 9ms/step Epoch 26/500 Epoch 00026: val_loss improved from 0.00477 to 0.00473, saving model to LSTM6.h5 17/17 - 0s - loss: 9.5968e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 129ms/epoch - 8ms/step Epoch 27/500 Epoch 00027: val_loss improved from 0.00473 to 0.00470, saving model to LSTM6.h5 17/17 - 0s - loss: 9.5080e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 136ms/epoch - 8ms/step Epoch 28/500 Epoch 00028: val_loss improved from 0.00470 to 0.00467, saving model to LSTM6.h5 17/17 - 0s - loss: 9.4306e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 134ms/epoch - 8ms/step Epoch 29/500 Epoch 00029: val_loss improved from 0.00467 to 0.00466, saving model to LSTM6.h5 17/17 - 0s - loss: 9.3621e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 128ms/epoch - 8ms/step Epoch 30/500 Epoch 00030: val_loss improved from 0.00466 to 0.00465, saving model to LSTM6.h5 17/17 - 0s - loss: 9.3003e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 136ms/epoch - 8ms/step Epoch 31/500 Epoch 00031: val_loss improved from 0.00465 to 0.00465, saving model to LSTM6.h5 17/17 - 0s - loss: 9.2438e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 139ms/epoch - 8ms/step Epoch 32/500 Epoch 00032: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00032: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.1913e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 122ms/epoch - 7ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0969e-04 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0925e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 116ms/epoch - 7ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0880e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 123ms/epoch - 7ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0836e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 117ms/epoch - 7ms/step Epoch 37/500 Epoch 00037: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00037: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0790e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0744e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 109ms/epoch - 6ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0697e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 107ms/epoch - 6ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0650e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 107ms/epoch - 6ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0602e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0553e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 120ms/epoch - 7ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0504e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 114ms/epoch - 7ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0454e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 109ms/epoch - 6ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0404e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0354e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 114ms/epoch - 7ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0302e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 112ms/epoch - 7ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0251e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 108ms/epoch - 6ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0198e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 124ms/epoch - 7ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0146e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0093e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 113ms/epoch - 7ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.00465 17/17 - 0s - loss: 9.0039e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 108ms/epoch - 6ms/step Epoch 53/500 Epoch 00053: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9985e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 107ms/epoch - 6ms/step Epoch 54/500 Epoch 00054: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9931e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 114ms/epoch - 7ms/step Epoch 55/500 Epoch 00055: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9876e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 122ms/epoch - 7ms/step Epoch 56/500 Epoch 00056: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9821e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 131ms/epoch - 8ms/step Epoch 57/500 Epoch 00057: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9766e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 165ms/epoch - 10ms/step Epoch 58/500 Epoch 00058: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9710e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 113ms/epoch - 7ms/step Epoch 59/500 Epoch 00059: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9654e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 127ms/epoch - 7ms/step Epoch 60/500 Epoch 00060: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9597e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step Epoch 61/500 Epoch 00061: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9540e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 116ms/epoch - 7ms/step Epoch 62/500 Epoch 00062: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9483e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 116ms/epoch - 7ms/step Epoch 63/500 Epoch 00063: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9425e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 117ms/epoch - 7ms/step Epoch 64/500 Epoch 00064: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9367e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 125ms/epoch - 7ms/step Epoch 65/500 Epoch 00065: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9309e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 115ms/epoch - 7ms/step Epoch 66/500 Epoch 00066: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9250e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step Epoch 67/500 Epoch 00067: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9191e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 122ms/epoch - 7ms/step Epoch 68/500 Epoch 00068: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9132e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 128ms/epoch - 8ms/step Epoch 69/500 Epoch 00069: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9072e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 118ms/epoch - 7ms/step Epoch 70/500 Epoch 00070: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.9012e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step Epoch 71/500 Epoch 00071: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.8952e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 100ms/epoch - 6ms/step Epoch 72/500 Epoch 00072: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.8892e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 115ms/epoch - 7ms/step Epoch 73/500 Epoch 00073: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.8831e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 109ms/epoch - 6ms/step Epoch 74/500 Epoch 00074: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.8770e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 113ms/epoch - 7ms/step Epoch 75/500 Epoch 00075: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.8709e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 110ms/epoch - 6ms/step Epoch 76/500 Epoch 00076: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.8647e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 122ms/epoch - 7ms/step Epoch 77/500 Epoch 00077: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.8585e-04 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 120ms/epoch - 7ms/step Epoch 78/500 Epoch 00078: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.8523e-04 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 106ms/epoch - 6ms/step Epoch 79/500 Epoch 00079: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.8461e-04 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 118ms/epoch - 7ms/step Epoch 80/500 Epoch 00080: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.8398e-04 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 112ms/epoch - 7ms/step Epoch 81/500 Epoch 00081: val_loss did not improve from 0.00465 17/17 - 0s - loss: 8.8335e-04 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step Epoch 00081: early stopping
SMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 109.05489356902385
RMSE: 10.442935103170173
MAPE: 8.732625329283675
EMA
Prediction vs Close: 54.85% Accuracy
Prediction vs Prediction: 44.4% Accuracy
MSE: 73.04142380966745
RMSE: 8.546427546622475
MAPE: 7.099244401385842
WMA
Prediction vs Close: 53.36% Accuracy
Prediction vs Prediction: 44.78% Accuracy
MSE: 84.94978866341171
RMSE: 9.21682096296829
MAPE: 7.490547440692417
DEMA
DEMA([input_arrays], [timeperiod=30])
Double Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
89
Working on DEMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.785, Time=3.12 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.588, Time=4.69 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-15575.689, Time=9.23 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.588, Time=7.27 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16714.796, Time=9.13 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-15610.140, Time=10.56 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17225.835, Time=22.70 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-14562.588, Time=9.42 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-16751.951, Time=21.31 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-11788.089, Time=31.05 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 128.505 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8639.917
Date: Sun, 12 Dec 2021 AIC -17225.835
Time: 16:30:59 BIC -17099.182
Sample: 0 HQIC -17177.195
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -5.894e-09 3.61e-06 -0.002 0.999 -7.09e-06 7.08e-06
x2 -5.93e-09 3.63e-06 -0.002 0.999 -7.11e-06 7.1e-06
x3 -5.905e-09 3.62e-06 -0.002 0.999 -7.1e-06 7.09e-06
x4 1.0000 3.62e-06 2.76e+05 0.000 1.000 1.000
x5 -5.457e-09 3.48e-06 -0.002 0.999 -6.83e-06 6.82e-06
x6 -3.019e-08 7.72e-06 -0.004 0.997 -1.52e-05 1.51e-05
x7 -5.87e-09 3.61e-06 -0.002 0.999 -7.08e-06 7.07e-06
x8 -5.809e-09 3.59e-06 -0.002 0.999 -7.05e-06 7.04e-06
x9 -9.293e-11 9.83e-08 -0.001 0.999 -1.93e-07 1.93e-07
x10 -2.793e-09 2.47e-06 -0.001 0.999 -4.84e-06 4.84e-06
x11 -6.095e-09 3.68e-06 -0.002 0.999 -7.21e-06 7.2e-06
x12 -5.478e-09 3.49e-06 -0.002 0.999 -6.85e-06 6.84e-06
x13 -5.91e-09 3.62e-06 -0.002 0.999 -7.1e-06 7.09e-06
x14 -4.085e-08 9.35e-06 -0.004 0.997 -1.84e-05 1.83e-05
x15 -5.93e-09 3.63e-06 -0.002 0.999 -7.12e-06 7.11e-06
x16 -1.618e-09 1.92e-06 -0.001 0.999 -3.76e-06 3.75e-06
x17 -5.076e-09 3.37e-06 -0.002 0.999 -6.6e-06 6.59e-06
x18 -1.377e-08 5.5e-06 -0.003 0.998 -1.08e-05 1.08e-05
x19 -6.135e-09 3.69e-06 -0.002 0.999 -7.23e-06 7.22e-06
x20 -1.018e-08 4.43e-06 -0.002 0.998 -8.68e-06 8.66e-06
x21 -6.911e-08 1.21e-05 -0.006 0.995 -2.39e-05 2.37e-05
x22 -5.656e-08 1.1e-05 -0.005 0.996 -2.16e-05 2.15e-05
x23 -5.355e-08 1.07e-05 -0.005 0.996 -2.1e-05 2.09e-05
x24 -3.636e-08 8.85e-06 -0.004 0.997 -1.74e-05 1.73e-05
ma.L1 -1.3899 4.86e-11 -2.86e+10 0.000 -1.390 -1.390
ma.L2 0.4032 4.6e-11 8.76e+09 0.000 0.403 0.403
sigma2 7.526e-11 6.92e-11 1.088 0.277 -6.03e-11 2.11e-10
===================================================================================
Ljung-Box (L1) (Q): 69.65 Jarque-Bera (JB): 6422892.15
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 12.42
Prob(H) (two-sided): 0.00 Kurtosis: 439.89
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 2.74e+29. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
/usr/local/lib/python3.7/dist-packages/keras/optimizer_v2/adam.py:105: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead. super(Adam, self).__init__(name, **kwargs)
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.01647, saving model to LSTM6.h5 10/10 - 4s - loss: 0.2634 - accuracy: 0.0000e+00 - val_loss: 0.0165 - val_accuracy: 0.0037 - lr: 0.0010 - 4s/epoch - 397ms/step Epoch 2/500 Epoch 00002: val_loss did not improve from 0.01647 10/10 - 0s - loss: 0.1891 - accuracy: 0.0000e+00 - val_loss: 0.0248 - val_accuracy: 0.0037 - lr: 0.0010 - 70ms/epoch - 7ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.01647 10/10 - 0s - loss: 0.0093 - accuracy: 0.0000e+00 - val_loss: 0.0261 - val_accuracy: 0.0037 - lr: 0.0010 - 77ms/epoch - 8ms/step Epoch 4/500 Epoch 00004: val_loss did not improve from 0.01647 10/10 - 0s - loss: 0.0087 - accuracy: 0.0000e+00 - val_loss: 0.0570 - val_accuracy: 0.0037 - lr: 0.0010 - 78ms/epoch - 8ms/step Epoch 5/500 Epoch 00005: val_loss improved from 0.01647 to 0.01335, saving model to LSTM6.h5 10/10 - 0s - loss: 0.0097 - accuracy: 0.0000e+00 - val_loss: 0.0134 - val_accuracy: 0.0037 - lr: 0.0010 - 88ms/epoch - 9ms/step Epoch 6/500 Epoch 00006: val_loss improved from 0.01335 to 0.01307, saving model to LSTM6.h5 10/10 - 0s - loss: 0.0028 - accuracy: 0.0000e+00 - val_loss: 0.0131 - val_accuracy: 0.0037 - lr: 0.0010 - 86ms/epoch - 9ms/step Epoch 7/500 Epoch 00007: val_loss improved from 0.01307 to 0.00753, saving model to LSTM6.h5 10/10 - 0s - loss: 0.0024 - accuracy: 0.0000e+00 - val_loss: 0.0075 - val_accuracy: 0.0037 - lr: 0.0010 - 93ms/epoch - 9ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.00753 10/10 - 0s - loss: 0.0017 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 0.0010 - 88ms/epoch - 9ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.00753 10/10 - 0s - loss: 0.0015 - accuracy: 0.0000e+00 - val_loss: 0.0130 - val_accuracy: 0.0037 - lr: 0.0010 - 80ms/epoch - 8ms/step Epoch 10/500 Epoch 00010: val_loss did not improve from 0.00753 10/10 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0168 - val_accuracy: 0.0037 - lr: 0.0010 - 66ms/epoch - 7ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.00753 10/10 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0118 - val_accuracy: 0.0037 - lr: 0.0010 - 67ms/epoch - 7ms/step Epoch 12/500 Epoch 00012: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00012: val_loss did not improve from 0.00753 10/10 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0156 - val_accuracy: 0.0037 - lr: 0.0010 - 85ms/epoch - 8ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.00753 10/10 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0151 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 72ms/epoch - 7ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.00753 10/10 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0144 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 79ms/epoch - 8ms/step Epoch 15/500 Epoch 00015: val_loss did not improve from 0.00753 10/10 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0141 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 88ms/epoch - 9ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.00753 10/10 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 71ms/epoch - 7ms/step Epoch 17/500 Epoch 00017: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00017: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.9688e-04 - accuracy: 0.0000e+00 - val_loss: 0.0144 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 70ms/epoch - 7ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.9189e-04 - accuracy: 0.0000e+00 - val_loss: 0.0144 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 70ms/epoch - 7ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.9169e-04 - accuracy: 0.0000e+00 - val_loss: 0.0144 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 68ms/epoch - 7ms/step Epoch 20/500 Epoch 00020: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.9147e-04 - accuracy: 0.0000e+00 - val_loss: 0.0144 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 76ms/epoch - 8ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.9125e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 74ms/epoch - 7ms/step Epoch 22/500 Epoch 00022: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00022: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.9101e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 75ms/epoch - 8ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.9077e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 75ms/epoch - 7ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.9051e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 74ms/epoch - 7ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.9025e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 88ms/epoch - 9ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8998e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 81ms/epoch - 8ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8970e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 100ms/epoch - 10ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8941e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 83ms/epoch - 8ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8911e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 96ms/epoch - 10ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8881e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 79ms/epoch - 8ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8850e-04 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 91ms/epoch - 9ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8817e-04 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 89ms/epoch - 9ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8785e-04 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 71ms/epoch - 7ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8751e-04 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 73ms/epoch - 7ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8717e-04 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 78ms/epoch - 8ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8682e-04 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 71ms/epoch - 7ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8647e-04 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 73ms/epoch - 7ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8610e-04 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 78ms/epoch - 8ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8574e-04 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 82ms/epoch - 8ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8537e-04 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 88ms/epoch - 9ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8499e-04 - accuracy: 0.0000e+00 - val_loss: 0.0147 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 85ms/epoch - 8ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8460e-04 - accuracy: 0.0000e+00 - val_loss: 0.0147 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 73ms/epoch - 7ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8422e-04 - accuracy: 0.0000e+00 - val_loss: 0.0147 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 71ms/epoch - 7ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8382e-04 - accuracy: 0.0000e+00 - val_loss: 0.0147 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 77ms/epoch - 8ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8342e-04 - accuracy: 0.0000e+00 - val_loss: 0.0147 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 87ms/epoch - 9ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8302e-04 - accuracy: 0.0000e+00 - val_loss: 0.0147 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 81ms/epoch - 8ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8261e-04 - accuracy: 0.0000e+00 - val_loss: 0.0147 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 83ms/epoch - 8ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8220e-04 - accuracy: 0.0000e+00 - val_loss: 0.0147 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 75ms/epoch - 8ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8178e-04 - accuracy: 0.0000e+00 - val_loss: 0.0147 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 75ms/epoch - 7ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8136e-04 - accuracy: 0.0000e+00 - val_loss: 0.0147 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 75ms/epoch - 8ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8094e-04 - accuracy: 0.0000e+00 - val_loss: 0.0148 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 106ms/epoch - 11ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8050e-04 - accuracy: 0.0000e+00 - val_loss: 0.0148 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 71ms/epoch - 7ms/step Epoch 53/500 Epoch 00053: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.8007e-04 - accuracy: 0.0000e+00 - val_loss: 0.0148 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 79ms/epoch - 8ms/step Epoch 54/500 Epoch 00054: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.7963e-04 - accuracy: 0.0000e+00 - val_loss: 0.0148 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 77ms/epoch - 8ms/step Epoch 55/500 Epoch 00055: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.7919e-04 - accuracy: 0.0000e+00 - val_loss: 0.0148 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 80ms/epoch - 8ms/step Epoch 56/500 Epoch 00056: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.7874e-04 - accuracy: 0.0000e+00 - val_loss: 0.0148 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 100ms/epoch - 10ms/step Epoch 57/500 Epoch 00057: val_loss did not improve from 0.00753 10/10 - 0s - loss: 9.7828e-04 - accuracy: 0.0000e+00 - val_loss: 0.0148 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 101ms/epoch - 10ms/step Epoch 00057: early stopping
SMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 109.05489356902385
RMSE: 10.442935103170173
MAPE: 8.732625329283675
EMA
Prediction vs Close: 54.85% Accuracy
Prediction vs Prediction: 44.4% Accuracy
MSE: 73.04142380966745
RMSE: 8.546427546622475
MAPE: 7.099244401385842
WMA
Prediction vs Close: 53.36% Accuracy
Prediction vs Prediction: 44.78% Accuracy
MSE: 84.94978866341171
RMSE: 9.21682096296829
MAPE: 7.490547440692417
DEMA
Prediction vs Close: 52.24% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 151.61070955572364
RMSE: 12.313030072070955
MAPE: 11.085595013418024
KAMA
KAMA([input_arrays], [timeperiod=30])
Kaufman Adaptive Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
18
Working on KAMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16921.943, Time=10.24 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.592, Time=4.52 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-16797.275, Time=9.03 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.592, Time=6.83 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16996.465, Time=3.42 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=-16999.509, Time=3.17 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=-17171.315, Time=6.58 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=-16994.523, Time=3.65 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=-15518.026, Time=29.06 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 76.513 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood 8613.658
Date: Sun, 12 Dec 2021 AIC -17171.315
Time: 16:36:46 BIC -17039.972
Sample: 0 HQIC -17120.874
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -5.14e-10 7.6e-05 -6.76e-06 1.000 -0.000 0.000
x2 -5.041e-10 7.52e-05 -6.7e-06 1.000 -0.000 0.000
x3 -4.834e-10 7.38e-05 -6.55e-06 1.000 -0.000 0.000
x4 1.0000 7.46e-05 1.34e+04 0.000 1.000 1.000
x5 -4.462e-10 7.09e-05 -6.29e-06 1.000 -0.000 0.000
x6 -3.064e-09 0.000 -1.84e-05 1.000 -0.000 0.000
x7 -4.751e-10 7.35e-05 -6.46e-06 1.000 -0.000 0.000
x8 -4.628e-10 7.28e-05 -6.36e-06 1.000 -0.000 0.000
x9 -9.21e-11 9.37e-06 -9.83e-06 1.000 -1.84e-05 1.84e-05
x10 -2.165e-10 3.1e-05 -6.98e-06 1.000 -6.08e-05 6.08e-05
x11 -4.665e-10 7.28e-05 -6.41e-06 1.000 -0.000 0.000
x12 -4.62e-10 7.23e-05 -6.39e-06 1.000 -0.000 0.000
x13 -4.906e-10 7.43e-05 -6.6e-06 1.000 -0.000 0.000
x14 -3.985e-09 0.000 -1.87e-05 1.000 -0.000 0.000
x15 -4.897e-10 7.48e-05 -6.55e-06 1.000 -0.000 0.000
x16 -7.327e-10 9.24e-05 -7.93e-06 1.000 -0.000 0.000
x17 -4.173e-10 6.93e-05 -6.02e-06 1.000 -0.000 0.000
x18 -3.397e-10 6.02e-05 -5.64e-06 1.000 -0.000 0.000
x19 -6.012e-10 8.3e-05 -7.25e-06 1.000 -0.000 0.000
x20 -9.09e-10 0.000 -9.05e-06 1.000 -0.000 0.000
x21 -6.188e-09 0.000 -2.32e-05 1.000 -0.001 0.001
x22 -1.992e-09 0.000 -1.33e-05 1.000 -0.000 0.000
x23 -3.669e-09 0.000 -1.79e-05 1.000 -0.000 0.000
x24 -1.065e-09 0.000 -1.01e-05 1.000 -0.000 0.000
ar.L1 -1.2073 5.73e-10 -2.11e+09 0.000 -1.207 -1.207
ar.L2 -0.9083 5.93e-10 -1.53e+09 0.000 -0.908 -0.908
ar.L3 -0.4033 5.84e-10 -6.91e+08 0.000 -0.403 -0.403
sigma2 8.06e-11 6.94e-11 1.162 0.245 -5.54e-11 2.17e-10
===================================================================================
Ljung-Box (L1) (Q): 13.77 Jarque-Bera (JB): 2436796.68
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 4.07
Prob(H) (two-sided): 0.00 Kurtosis: 272.41
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.23e+28. Standard errors may be unstable.
ARIMA order: (3, 3, 0)
/usr/local/lib/python3.7/dist-packages/keras/optimizer_v2/adam.py:105: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead. super(Adam, self).__init__(name, **kwargs)
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.10479, saving model to LSTM6.h5 45/45 - 5s - loss: 0.1250 - accuracy: 0.0000e+00 - val_loss: 0.1048 - val_accuracy: 0.0000e+00 - lr: 0.0010 - 5s/epoch - 102ms/step Epoch 2/500 Epoch 00002: val_loss improved from 0.10479 to 0.08778, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0877 - accuracy: 0.0000e+00 - val_loss: 0.0878 - val_accuracy: 0.0037 - lr: 0.0010 - 295ms/epoch - 7ms/step Epoch 3/500 Epoch 00003: val_loss improved from 0.08778 to 0.00637, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0120 - accuracy: 0.0000e+00 - val_loss: 0.0064 - val_accuracy: 0.0037 - lr: 0.0010 - 308ms/epoch - 7ms/step Epoch 4/500 Epoch 00004: val_loss did not improve from 0.00637 45/45 - 0s - loss: 0.0063 - accuracy: 0.0000e+00 - val_loss: 0.0347 - val_accuracy: 0.0037 - lr: 0.0010 - 295ms/epoch - 7ms/step Epoch 5/500 Epoch 00005: val_loss improved from 0.00637 to 0.00356, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0064 - accuracy: 0.0000e+00 - val_loss: 0.0036 - val_accuracy: 0.0037 - lr: 0.0010 - 263ms/epoch - 6ms/step Epoch 6/500 Epoch 00006: val_loss did not improve from 0.00356 45/45 - 0s - loss: 0.0023 - accuracy: 0.0000e+00 - val_loss: 0.0121 - val_accuracy: 0.0037 - lr: 0.0010 - 259ms/epoch - 6ms/step Epoch 7/500 Epoch 00007: val_loss did not improve from 0.00356 45/45 - 0s - loss: 0.0017 - accuracy: 0.0000e+00 - val_loss: 0.0039 - val_accuracy: 0.0037 - lr: 0.0010 - 262ms/epoch - 6ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.00356 45/45 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 0.0010 - 264ms/epoch - 6ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.00356 45/45 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0037 - val_accuracy: 0.0037 - lr: 0.0010 - 244ms/epoch - 5ms/step Epoch 10/500 Epoch 00010: val_loss improved from 0.00356 to 0.00305, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0030 - val_accuracy: 0.0037 - lr: 0.0010 - 289ms/epoch - 6ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0078 - val_accuracy: 0.0037 - lr: 0.0010 - 250ms/epoch - 6ms/step Epoch 12/500 Epoch 00012: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0065 - val_accuracy: 0.0037 - lr: 0.0010 - 240ms/epoch - 5ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0026 - accuracy: 0.0000e+00 - val_loss: 0.0114 - val_accuracy: 0.0037 - lr: 0.0010 - 269ms/epoch - 6ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0064 - accuracy: 0.0000e+00 - val_loss: 0.0091 - val_accuracy: 0.0037 - lr: 0.0010 - 286ms/epoch - 6ms/step Epoch 15/500 Epoch 00015: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00015: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0176 - accuracy: 0.0000e+00 - val_loss: 0.0089 - val_accuracy: 0.0037 - lr: 0.0010 - 271ms/epoch - 6ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0164 - accuracy: 0.0000e+00 - val_loss: 0.0405 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 253ms/epoch - 6ms/step Epoch 17/500 Epoch 00017: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0120 - accuracy: 0.0000e+00 - val_loss: 0.0169 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 258ms/epoch - 6ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0059 - accuracy: 0.0000e+00 - val_loss: 0.0108 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 268ms/epoch - 6ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0042 - accuracy: 0.0000e+00 - val_loss: 0.0071 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 280ms/epoch - 6ms/step Epoch 20/500 Epoch 00020: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00020: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0032 - accuracy: 0.0000e+00 - val_loss: 0.0051 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 247ms/epoch - 5ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0022 - accuracy: 0.0000e+00 - val_loss: 0.0049 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 251ms/epoch - 6ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0021 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 250ms/epoch - 6ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0021 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 277ms/epoch - 6ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0020 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 257ms/epoch - 6ms/step Epoch 25/500 Epoch 00025: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00025: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0020 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 242ms/epoch - 5ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0020 - accuracy: 0.0000e+00 - val_loss: 0.0043 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 271ms/epoch - 6ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0019 - accuracy: 0.0000e+00 - val_loss: 0.0042 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 261ms/epoch - 6ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0019 - accuracy: 0.0000e+00 - val_loss: 0.0042 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 268ms/epoch - 6ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0019 - accuracy: 0.0000e+00 - val_loss: 0.0041 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 266ms/epoch - 6ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0018 - accuracy: 0.0000e+00 - val_loss: 0.0040 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 239ms/epoch - 5ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0018 - accuracy: 0.0000e+00 - val_loss: 0.0039 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 234ms/epoch - 5ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0018 - accuracy: 0.0000e+00 - val_loss: 0.0038 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 256ms/epoch - 6ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0017 - accuracy: 0.0000e+00 - val_loss: 0.0038 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 272ms/epoch - 6ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0017 - accuracy: 0.0000e+00 - val_loss: 0.0037 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 252ms/epoch - 6ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0017 - accuracy: 0.0000e+00 - val_loss: 0.0036 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 279ms/epoch - 6ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0017 - accuracy: 0.0000e+00 - val_loss: 0.0035 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 289ms/epoch - 6ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0016 - accuracy: 0.0000e+00 - val_loss: 0.0035 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 317ms/epoch - 7ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0016 - accuracy: 0.0000e+00 - val_loss: 0.0034 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 263ms/epoch - 6ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0016 - accuracy: 0.0000e+00 - val_loss: 0.0033 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 250ms/epoch - 6ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0016 - accuracy: 0.0000e+00 - val_loss: 0.0033 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 270ms/epoch - 6ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0015 - accuracy: 0.0000e+00 - val_loss: 0.0032 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 240ms/epoch - 5ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0015 - accuracy: 0.0000e+00 - val_loss: 0.0032 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 237ms/epoch - 5ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0015 - accuracy: 0.0000e+00 - val_loss: 0.0031 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 289ms/epoch - 6ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.00305 45/45 - 0s - loss: 0.0015 - accuracy: 0.0000e+00 - val_loss: 0.0031 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 248ms/epoch - 6ms/step Epoch 45/500 Epoch 00045: val_loss improved from 0.00305 to 0.00302, saving model to LSTM6.h5 45/45 - 1s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0030 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 728ms/epoch - 16ms/step Epoch 46/500 Epoch 00046: val_loss improved from 0.00302 to 0.00298, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0030 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 274ms/epoch - 6ms/step Epoch 47/500 Epoch 00047: val_loss improved from 0.00298 to 0.00294, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 306ms/epoch - 7ms/step Epoch 48/500 Epoch 00048: val_loss improved from 0.00294 to 0.00291, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 280ms/epoch - 6ms/step Epoch 49/500 Epoch 00049: val_loss improved from 0.00291 to 0.00288, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0014 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 309ms/epoch - 7ms/step Epoch 50/500 Epoch 00050: val_loss improved from 0.00288 to 0.00285, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 297ms/epoch - 7ms/step Epoch 51/500 Epoch 00051: val_loss improved from 0.00285 to 0.00282, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 283ms/epoch - 6ms/step Epoch 52/500 Epoch 00052: val_loss improved from 0.00282 to 0.00280, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 294ms/epoch - 7ms/step Epoch 53/500 Epoch 00053: val_loss improved from 0.00280 to 0.00278, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 288ms/epoch - 6ms/step Epoch 54/500 Epoch 00054: val_loss improved from 0.00278 to 0.00276, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 261ms/epoch - 6ms/step Epoch 55/500 Epoch 00055: val_loss improved from 0.00276 to 0.00274, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0027 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 271ms/epoch - 6ms/step Epoch 56/500 Epoch 00056: val_loss improved from 0.00274 to 0.00273, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0027 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 283ms/epoch - 6ms/step Epoch 57/500 Epoch 00057: val_loss improved from 0.00273 to 0.00272, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0027 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 309ms/epoch - 7ms/step Epoch 58/500 Epoch 00058: val_loss improved from 0.00272 to 0.00272, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0027 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 321ms/epoch - 7ms/step Epoch 59/500 Epoch 00059: val_loss improved from 0.00272 to 0.00271, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0027 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 295ms/epoch - 7ms/step Epoch 60/500 Epoch 00060: val_loss improved from 0.00271 to 0.00271, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0027 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 293ms/epoch - 7ms/step Epoch 61/500 Epoch 00061: val_loss improved from 0.00271 to 0.00271, saving model to LSTM6.h5 45/45 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0027 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 289ms/epoch - 6ms/step Epoch 62/500 Epoch 00062: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0027 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 283ms/epoch - 6ms/step Epoch 63/500 Epoch 00063: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0027 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 274ms/epoch - 6ms/step Epoch 64/500 Epoch 00064: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0027 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 262ms/epoch - 6ms/step Epoch 65/500 Epoch 00065: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0027 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 270ms/epoch - 6ms/step Epoch 66/500 Epoch 00066: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0027 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 265ms/epoch - 6ms/step Epoch 67/500 Epoch 00067: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0027 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 256ms/epoch - 6ms/step Epoch 68/500 Epoch 00068: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0027 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 279ms/epoch - 6ms/step Epoch 69/500 Epoch 00069: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0027 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 243ms/epoch - 5ms/step Epoch 70/500 Epoch 00070: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 271ms/epoch - 6ms/step Epoch 71/500 Epoch 00071: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 278ms/epoch - 6ms/step Epoch 72/500 Epoch 00072: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 258ms/epoch - 6ms/step Epoch 73/500 Epoch 00073: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 288ms/epoch - 6ms/step Epoch 74/500 Epoch 00074: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 264ms/epoch - 6ms/step Epoch 75/500 Epoch 00075: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 239ms/epoch - 5ms/step Epoch 76/500 Epoch 00076: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 243ms/epoch - 5ms/step Epoch 77/500 Epoch 00077: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 292ms/epoch - 6ms/step Epoch 78/500 Epoch 00078: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 244ms/epoch - 5ms/step Epoch 79/500 Epoch 00079: val_loss did not improve from 0.00271 45/45 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 244ms/epoch - 5ms/step Epoch 80/500 Epoch 00080: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.9892e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 251ms/epoch - 6ms/step Epoch 81/500 Epoch 00081: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.9291e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 276ms/epoch - 6ms/step Epoch 82/500 Epoch 00082: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.8710e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 279ms/epoch - 6ms/step Epoch 83/500 Epoch 00083: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.8146e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 259ms/epoch - 6ms/step Epoch 84/500 Epoch 00084: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.7599e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 242ms/epoch - 5ms/step Epoch 85/500 Epoch 00085: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.7069e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 272ms/epoch - 6ms/step Epoch 86/500 Epoch 00086: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.6554e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 276ms/epoch - 6ms/step Epoch 87/500 Epoch 00087: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.6053e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 284ms/epoch - 6ms/step Epoch 88/500 Epoch 00088: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.5565e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 268ms/epoch - 6ms/step Epoch 89/500 Epoch 00089: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.5091e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 259ms/epoch - 6ms/step Epoch 90/500 Epoch 00090: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.4628e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 283ms/epoch - 6ms/step Epoch 91/500 Epoch 00091: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.4176e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 261ms/epoch - 6ms/step Epoch 92/500 Epoch 00092: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.3735e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 281ms/epoch - 6ms/step Epoch 93/500 Epoch 00093: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.3304e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 246ms/epoch - 5ms/step Epoch 94/500 Epoch 00094: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.2883e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 253ms/epoch - 6ms/step Epoch 95/500 Epoch 00095: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.2470e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 265ms/epoch - 6ms/step Epoch 96/500 Epoch 00096: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.2065e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 268ms/epoch - 6ms/step Epoch 97/500 Epoch 00097: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.1668e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 262ms/epoch - 6ms/step Epoch 98/500 Epoch 00098: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.1279e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 288ms/epoch - 6ms/step Epoch 99/500 Epoch 00099: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.0896e-04 - accuracy: 0.0000e+00 - val_loss: 0.0029 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 308ms/epoch - 7ms/step Epoch 100/500 Epoch 00100: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.0519e-04 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 274ms/epoch - 6ms/step Epoch 101/500 Epoch 00101: val_loss did not improve from 0.00271 45/45 - 0s - loss: 9.0149e-04 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 292ms/epoch - 6ms/step Epoch 102/500 Epoch 00102: val_loss did not improve from 0.00271 45/45 - 0s - loss: 8.9784e-04 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 302ms/epoch - 7ms/step Epoch 103/500 Epoch 00103: val_loss did not improve from 0.00271 45/45 - 0s - loss: 8.9425e-04 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 242ms/epoch - 5ms/step Epoch 104/500 Epoch 00104: val_loss did not improve from 0.00271 45/45 - 0s - loss: 8.9070e-04 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 247ms/epoch - 5ms/step Epoch 105/500 Epoch 00105: val_loss did not improve from 0.00271 45/45 - 0s - loss: 8.8720e-04 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 281ms/epoch - 6ms/step Epoch 106/500 Epoch 00106: val_loss did not improve from 0.00271 45/45 - 0s - loss: 8.8375e-04 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 313ms/epoch - 7ms/step Epoch 107/500 Epoch 00107: val_loss did not improve from 0.00271 45/45 - 0s - loss: 8.8034e-04 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 261ms/epoch - 6ms/step Epoch 108/500 Epoch 00108: val_loss did not improve from 0.00271 45/45 - 0s - loss: 8.7696e-04 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 243ms/epoch - 5ms/step Epoch 109/500 Epoch 00109: val_loss did not improve from 0.00271 45/45 - 0s - loss: 8.7363e-04 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 255ms/epoch - 6ms/step Epoch 110/500 Epoch 00110: val_loss did not improve from 0.00271 45/45 - 0s - loss: 8.7033e-04 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 269ms/epoch - 6ms/step Epoch 111/500 Epoch 00111: val_loss did not improve from 0.00271 45/45 - 0s - loss: 8.6706e-04 - accuracy: 0.0000e+00 - val_loss: 0.0028 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 283ms/epoch - 6ms/step Epoch 00111: early stopping
SMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 109.05489356902385
RMSE: 10.442935103170173
MAPE: 8.732625329283675
EMA
Prediction vs Close: 54.85% Accuracy
Prediction vs Prediction: 44.4% Accuracy
MSE: 73.04142380966745
RMSE: 8.546427546622475
MAPE: 7.099244401385842
WMA
Prediction vs Close: 53.36% Accuracy
Prediction vs Prediction: 44.78% Accuracy
MSE: 84.94978866341171
RMSE: 9.21682096296829
MAPE: 7.490547440692417
DEMA
Prediction vs Close: 52.24% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 151.61070955572364
RMSE: 12.313030072070955
MAPE: 11.085595013418024
KAMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 80.0014001509936
RMSE: 8.944350180476702
MAPE: 7.358601729961256
MIDPOINT
MIDPOINT([input_arrays], [timeperiod=14])
MidPoint over period (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 14
Outputs:
real
14
Working on MIDPOINT predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.768, Time=3.03 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.591, Time=4.53 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-15581.065, Time=8.89 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.591, Time=7.32 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16536.628, Time=10.00 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-13971.493, Time=10.04 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17226.044, Time=20.24 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-14562.591, Time=9.42 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-16754.945, Time=19.03 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-15001.855, Time=20.62 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 113.134 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8640.022
Date: Sun, 12 Dec 2021 AIC -17226.044
Time: 16:40:50 BIC -17099.391
Sample: 0 HQIC -17177.404
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -5.031e-09 1.06e-05 -0.000 1.000 -2.08e-05 2.08e-05
x2 -4.99e-09 8.12e-06 -0.001 1.000 -1.59e-05 1.59e-05
x3 -5.114e-09 1.38e-05 -0.000 1.000 -2.7e-05 2.7e-05
x4 1.0000 8.91e-06 1.12e+05 0.000 1.000 1.000
x5 -4.55e-09 8.2e-06 -0.001 1.000 -1.61e-05 1.61e-05
x6 -9.992e-08 0.001 -0.000 1.000 -0.002 0.002
x7 -4.607e-09 1.97e-05 -0.000 1.000 -3.86e-05 3.86e-05
x8 -4.591e-09 1.77e-05 -0.000 1.000 -3.48e-05 3.48e-05
x9 -2.538e-09 1.13e-05 -0.000 1.000 -2.21e-05 2.21e-05
x10 -4.315e-09 6.08e-06 -0.001 0.999 -1.19e-05 1.19e-05
x11 -4.545e-09 1.62e-05 -0.000 1.000 -3.18e-05 3.18e-05
x12 -4.701e-09 1.97e-05 -0.000 1.000 -3.87e-05 3.87e-05
x13 -4.823e-09 1.18e-05 -0.000 1.000 -2.3e-05 2.3e-05
x14 -4.08e-08 4.99e-05 -0.001 0.999 -9.79e-05 9.78e-05
x15 -5.557e-09 2.03e-05 -0.000 1.000 -3.99e-05 3.99e-05
x16 -3.541e-09 1.3e-05 -0.000 1.000 -2.55e-05 2.55e-05
x17 -3.463e-09 1.51e-05 -0.000 1.000 -2.97e-05 2.97e-05
x18 -1.534e-08 4e-05 -0.000 1.000 -7.85e-05 7.85e-05
x19 -6.118e-09 2.07e-05 -0.000 1.000 -4.05e-05 4.05e-05
x20 -1.581e-08 3.38e-05 -0.000 1.000 -6.62e-05 6.61e-05
x21 -5.505e-08 5.6e-05 -0.001 0.999 -0.000 0.000
x22 -2.936e-08 4.55e-05 -0.001 0.999 -8.92e-05 8.92e-05
x23 -3.882e-08 4.89e-05 -0.001 0.999 -9.58e-05 9.57e-05
x24 -2.099e-08 4.87e-05 -0.000 1.000 -9.54e-05 9.54e-05
ma.L1 -1.3900 1.23e-07 -1.13e+07 0.000 -1.390 -1.390
ma.L2 0.4044 1.43e-07 2.82e+06 0.000 0.404 0.404
sigma2 7.525e-11 7.22e-11 1.042 0.297 -6.63e-11 2.17e-10
===================================================================================
Ljung-Box (L1) (Q): 55.84 Jarque-Bera (JB): 1335305.59
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.09 Skew: 5.74
Prob(H) (two-sided): 0.00 Kurtosis: 202.19
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.77e+23. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
/usr/local/lib/python3.7/dist-packages/keras/optimizer_v2/adam.py:105: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead. super(Adam, self).__init__(name, **kwargs)
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.03788, saving model to LSTM6.h5 58/58 - 4s - loss: 0.1476 - accuracy: 0.0000e+00 - val_loss: 0.0379 - val_accuracy: 0.0037 - lr: 0.0010 - 4s/epoch - 68ms/step Epoch 2/500 Epoch 00002: val_loss improved from 0.03788 to 0.00696, saving model to LSTM6.h5 58/58 - 0s - loss: 0.0220 - accuracy: 0.0000e+00 - val_loss: 0.0070 - val_accuracy: 0.0037 - lr: 0.0010 - 347ms/epoch - 6ms/step Epoch 3/500 Epoch 00003: val_loss did not improve from 0.00696 58/58 - 0s - loss: 0.0223 - accuracy: 0.0000e+00 - val_loss: 0.0802 - val_accuracy: 0.0037 - lr: 0.0010 - 352ms/epoch - 6ms/step Epoch 4/500 Epoch 00004: val_loss did not improve from 0.00696 58/58 - 0s - loss: 0.0103 - accuracy: 0.0000e+00 - val_loss: 0.0215 - val_accuracy: 0.0037 - lr: 0.0010 - 322ms/epoch - 6ms/step Epoch 5/500 Epoch 00005: val_loss did not improve from 0.00696 58/58 - 0s - loss: 0.0040 - accuracy: 0.0000e+00 - val_loss: 0.0176 - val_accuracy: 0.0037 - lr: 0.0010 - 331ms/epoch - 6ms/step Epoch 6/500 Epoch 00006: val_loss did not improve from 0.00696 58/58 - 0s - loss: 0.0066 - accuracy: 0.0000e+00 - val_loss: 0.0102 - val_accuracy: 0.0037 - lr: 0.0010 - 369ms/epoch - 6ms/step Epoch 7/500 Epoch 00007: val_loss improved from 0.00696 to 0.00664, saving model to LSTM6.h5 58/58 - 0s - loss: 0.0061 - accuracy: 0.0000e+00 - val_loss: 0.0066 - val_accuracy: 0.0037 - lr: 0.0010 - 346ms/epoch - 6ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.00664 58/58 - 0s - loss: 0.0104 - accuracy: 0.0000e+00 - val_loss: 0.0081 - val_accuracy: 0.0037 - lr: 0.0010 - 315ms/epoch - 5ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.00664 58/58 - 0s - loss: 0.0191 - accuracy: 0.0000e+00 - val_loss: 0.0079 - val_accuracy: 0.0037 - lr: 0.0010 - 330ms/epoch - 6ms/step Epoch 10/500 Epoch 00010: val_loss improved from 0.00664 to 0.00656, saving model to LSTM6.h5 58/58 - 0s - loss: 0.0280 - accuracy: 0.0000e+00 - val_loss: 0.0066 - val_accuracy: 0.0037 - lr: 0.0010 - 326ms/epoch - 6ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.00656 58/58 - 0s - loss: 0.0214 - accuracy: 0.0000e+00 - val_loss: 0.0112 - val_accuracy: 0.0037 - lr: 0.0010 - 346ms/epoch - 6ms/step Epoch 12/500 Epoch 00012: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00012: val_loss did not improve from 0.00656 58/58 - 0s - loss: 0.0049 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 0.0010 - 357ms/epoch - 6ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.00656 58/58 - 0s - loss: 0.0088 - accuracy: 0.0000e+00 - val_loss: 0.0075 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 314ms/epoch - 5ms/step Epoch 14/500 Epoch 00014: val_loss improved from 0.00656 to 0.00435, saving model to LSTM6.h5 58/58 - 0s - loss: 0.0060 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 345ms/epoch - 6ms/step Epoch 15/500 Epoch 00015: val_loss improved from 0.00435 to 0.00367, saving model to LSTM6.h5 58/58 - 0s - loss: 0.0041 - accuracy: 0.0000e+00 - val_loss: 0.0037 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 364ms/epoch - 6ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0031 - accuracy: 0.0000e+00 - val_loss: 0.0037 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 302ms/epoch - 5ms/step Epoch 17/500 Epoch 00017: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0024 - accuracy: 0.0000e+00 - val_loss: 0.0039 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 314ms/epoch - 5ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0020 - accuracy: 0.0000e+00 - val_loss: 0.0040 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 354ms/epoch - 6ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0017 - accuracy: 0.0000e+00 - val_loss: 0.0042 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 317ms/epoch - 5ms/step Epoch 20/500 Epoch 00020: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00020: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0015 - accuracy: 0.0000e+00 - val_loss: 0.0043 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 355ms/epoch - 6ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0044 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 354ms/epoch - 6ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0045 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 311ms/epoch - 5ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0046 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 326ms/epoch - 6ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 327ms/epoch - 6ms/step Epoch 25/500 Epoch 00025: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00025: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 294ms/epoch - 5ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 301ms/epoch - 5ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 330ms/epoch - 6ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 302ms/epoch - 5ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 302ms/epoch - 5ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 323ms/epoch - 6ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0048 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 336ms/epoch - 6ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0049 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 310ms/epoch - 5ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0049 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 332ms/epoch - 6ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0049 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 355ms/epoch - 6ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0049 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 306ms/epoch - 5ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0049 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 312ms/epoch - 5ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0050 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 372ms/epoch - 6ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0050 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 329ms/epoch - 6ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0050 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 332ms/epoch - 6ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0050 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 331ms/epoch - 6ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0051 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 318ms/epoch - 5ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.00367 58/58 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0051 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 324ms/epoch - 6ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.00367 58/58 - 0s - loss: 9.9583e-04 - accuracy: 0.0000e+00 - val_loss: 0.0051 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 313ms/epoch - 5ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.00367 58/58 - 0s - loss: 9.8837e-04 - accuracy: 0.0000e+00 - val_loss: 0.0052 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 304ms/epoch - 5ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.00367 58/58 - 0s - loss: 9.8109e-04 - accuracy: 0.0000e+00 - val_loss: 0.0052 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 336ms/epoch - 6ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.00367 58/58 - 0s - loss: 9.7399e-04 - accuracy: 0.0000e+00 - val_loss: 0.0052 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 334ms/epoch - 6ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.00367 58/58 - 0s - loss: 9.6709e-04 - accuracy: 0.0000e+00 - val_loss: 0.0053 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 325ms/epoch - 6ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.00367 58/58 - 0s - loss: 9.6037e-04 - accuracy: 0.0000e+00 - val_loss: 0.0053 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 320ms/epoch - 6ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.00367 58/58 - 0s - loss: 9.5385e-04 - accuracy: 0.0000e+00 - val_loss: 0.0053 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 313ms/epoch - 5ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.00367 58/58 - 0s - loss: 9.4751e-04 - accuracy: 0.0000e+00 - val_loss: 0.0054 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 313ms/epoch - 5ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.00367 58/58 - 0s - loss: 9.4136e-04 - accuracy: 0.0000e+00 - val_loss: 0.0054 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 318ms/epoch - 5ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.00367 58/58 - 0s - loss: 9.3539e-04 - accuracy: 0.0000e+00 - val_loss: 0.0055 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 340ms/epoch - 6ms/step Epoch 53/500 Epoch 00053: val_loss did not improve from 0.00367 58/58 - 0s - loss: 9.2961e-04 - accuracy: 0.0000e+00 - val_loss: 0.0055 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 317ms/epoch - 5ms/step Epoch 54/500 Epoch 00054: val_loss did not improve from 0.00367 58/58 - 0s - loss: 9.2400e-04 - accuracy: 0.0000e+00 - val_loss: 0.0055 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 304ms/epoch - 5ms/step Epoch 55/500 Epoch 00055: val_loss did not improve from 0.00367 58/58 - 0s - loss: 9.1856e-04 - accuracy: 0.0000e+00 - val_loss: 0.0056 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 322ms/epoch - 6ms/step Epoch 56/500 Epoch 00056: val_loss did not improve from 0.00367 58/58 - 0s - loss: 9.1329e-04 - accuracy: 0.0000e+00 - val_loss: 0.0056 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 306ms/epoch - 5ms/step Epoch 57/500 Epoch 00057: val_loss did not improve from 0.00367 58/58 - 0s - loss: 9.0818e-04 - accuracy: 0.0000e+00 - val_loss: 0.0057 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 338ms/epoch - 6ms/step Epoch 58/500 Epoch 00058: val_loss did not improve from 0.00367 58/58 - 0s - loss: 9.0322e-04 - accuracy: 0.0000e+00 - val_loss: 0.0057 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 332ms/epoch - 6ms/step Epoch 59/500 Epoch 00059: val_loss did not improve from 0.00367 58/58 - 0s - loss: 8.9842e-04 - accuracy: 0.0000e+00 - val_loss: 0.0058 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 324ms/epoch - 6ms/step Epoch 60/500 Epoch 00060: val_loss did not improve from 0.00367 58/58 - 0s - loss: 8.9375e-04 - accuracy: 0.0000e+00 - val_loss: 0.0058 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 342ms/epoch - 6ms/step Epoch 61/500 Epoch 00061: val_loss did not improve from 0.00367 58/58 - 0s - loss: 8.8922e-04 - accuracy: 0.0000e+00 - val_loss: 0.0058 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 337ms/epoch - 6ms/step Epoch 62/500 Epoch 00062: val_loss did not improve from 0.00367 58/58 - 0s - loss: 8.8481e-04 - accuracy: 0.0000e+00 - val_loss: 0.0059 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 334ms/epoch - 6ms/step Epoch 63/500 Epoch 00063: val_loss did not improve from 0.00367 58/58 - 0s - loss: 8.8053e-04 - accuracy: 0.0000e+00 - val_loss: 0.0059 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 329ms/epoch - 6ms/step Epoch 64/500 Epoch 00064: val_loss did not improve from 0.00367 58/58 - 0s - loss: 8.7636e-04 - accuracy: 0.0000e+00 - val_loss: 0.0060 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 364ms/epoch - 6ms/step Epoch 65/500 Epoch 00065: val_loss did not improve from 0.00367 58/58 - 0s - loss: 8.7229e-04 - accuracy: 0.0000e+00 - val_loss: 0.0060 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 326ms/epoch - 6ms/step Epoch 00065: early stopping
SMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 109.05489356902385
RMSE: 10.442935103170173
MAPE: 8.732625329283675
EMA
Prediction vs Close: 54.85% Accuracy
Prediction vs Prediction: 44.4% Accuracy
MSE: 73.04142380966745
RMSE: 8.546427546622475
MAPE: 7.099244401385842
WMA
Prediction vs Close: 53.36% Accuracy
Prediction vs Prediction: 44.78% Accuracy
MSE: 84.94978866341171
RMSE: 9.21682096296829
MAPE: 7.490547440692417
DEMA
Prediction vs Close: 52.24% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 151.61070955572364
RMSE: 12.313030072070955
MAPE: 11.085595013418024
KAMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 80.0014001509936
RMSE: 8.944350180476702
MAPE: 7.358601729961256
MIDPOINT
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 43.66% Accuracy
MSE: 61.18379984959283
RMSE: 7.822007405365507
MAPE: 6.441728946960992
T3
T3([input_arrays], [timeperiod=5], [vfactor=0.7])
Triple Exponential Moving Average (T3) (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 5
vfactor: 0.7
Outputs:
real
19
Working on T3 predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-17000.569, Time=3.20 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-15576.554, Time=5.82 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-16078.305, Time=8.22 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-15574.554, Time=9.60 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16998.627, Time=3.46 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-16429.916, Time=12.37 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17000.664, Time=3.29 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-15700.026, Time=11.25 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-15704.282, Time=15.04 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-16998.664, Time=3.21 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 75.481 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8527.332
Date: Sun, 12 Dec 2021 AIC -17000.664
Time: 16:46:57 BIC -16874.011
Sample: 0 HQIC -16952.024
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 8.378e-14 2.16e-06 3.89e-08 1.000 -4.23e-06 4.23e-06
x2 7.457e-14 2.15e-06 3.47e-08 1.000 -4.22e-06 4.22e-06
x3 2.279e-14 2.16e-06 1.05e-08 1.000 -4.24e-06 4.24e-06
x4 1.0000 2.16e-06 4.63e+05 0.000 1.000 1.000
x5 1.211e-12 2.07e-06 5.86e-07 1.000 -4.05e-06 4.05e-06
x6 3.146e-15 2.67e-06 1.18e-09 1.000 -5.23e-06 5.23e-06
x7 1.593e-13 2.15e-06 7.41e-08 1.000 -4.21e-06 4.21e-06
x8 -0.0001 2.1e-06 -48.778 0.000 -0.000 -9.82e-05
x9 5.141e-14 6.35e-07 8.1e-08 1.000 -1.24e-06 1.24e-06
x10 -6.174e-05 1.34e-06 -45.995 0.000 -6.44e-05 -5.91e-05
x11 0.0003 2.15e-06 148.354 0.000 0.000 0.000
x12 -0.0002 2.02e-06 -93.730 0.000 -0.000 -0.000
x13 1.967e-14 2.16e-06 9.1e-09 1.000 -4.23e-06 4.23e-06
x14 -1.297e-14 5.65e-06 -2.29e-09 1.000 -1.11e-05 1.11e-05
x15 -3.18e-12 1.82e-06 -1.75e-06 1.000 -3.57e-06 3.57e-06
x16 -1.426e-12 4.51e-06 -3.16e-07 1.000 -8.84e-06 8.84e-06
x17 7.474e-13 2.37e-06 3.16e-07 1.000 -4.64e-06 4.64e-06
x18 -2.92e-13 2.9e-06 -1.01e-07 1.000 -5.68e-06 5.68e-06
x19 -4.211e-14 1.89e-06 -2.22e-08 1.000 -3.71e-06 3.71e-06
x20 -1.515e-13 1.2e-06 -1.26e-07 1.000 -2.36e-06 2.36e-06
x21 6.555e-13 6.37e-06 1.03e-07 1.000 -1.25e-05 1.25e-05
x22 1.212e-14 6.19e-06 1.96e-09 1.000 -1.21e-05 1.21e-05
x23 -3.877e-13 3.76e-06 -1.03e-07 1.000 -7.38e-06 7.38e-06
x24 8.127e-15 4.01e-06 2.03e-09 1.000 -7.86e-06 7.86e-06
ma.L1 -1.3370 3.84e-12 -3.48e+11 0.000 -1.337 -1.337
ma.L2 0.4289 1.65e-12 2.6e+11 0.000 0.429 0.429
sigma2 1e-10 6.99e-11 1.430 0.153 -3.71e-11 2.37e-10
===================================================================================
Ljung-Box (L1) (Q): 4.57 Jarque-Bera (JB): 3228712.87
Prob(Q): 0.03 Prob(JB): 0.00
Heteroskedasticity (H): 0.12 Skew: -9.87
Prob(H) (two-sided): 0.00 Kurtosis: 312.63
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.57e+30. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
/usr/local/lib/python3.7/dist-packages/keras/optimizer_v2/adam.py:105: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead. super(Adam, self).__init__(name, **kwargs)
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.07461, saving model to LSTM6.h5 43/43 - 4s - loss: 0.1214 - accuracy: 0.0000e+00 - val_loss: 0.0746 - val_accuracy: 0.0037 - lr: 0.0010 - 4s/epoch - 104ms/step Epoch 2/500 Epoch 00002: val_loss improved from 0.07461 to 0.04371, saving model to LSTM6.h5 43/43 - 0s - loss: 0.0526 - accuracy: 0.0000e+00 - val_loss: 0.0437 - val_accuracy: 0.0037 - lr: 0.0010 - 291ms/epoch - 7ms/step Epoch 3/500 Epoch 00003: val_loss improved from 0.04371 to 0.00965, saving model to LSTM6.h5 43/43 - 0s - loss: 0.0168 - accuracy: 0.0000e+00 - val_loss: 0.0096 - val_accuracy: 0.0037 - lr: 0.0010 - 267ms/epoch - 6ms/step Epoch 4/500 Epoch 00004: val_loss did not improve from 0.00965 43/43 - 0s - loss: 0.0040 - accuracy: 0.0000e+00 - val_loss: 0.0155 - val_accuracy: 0.0037 - lr: 0.0010 - 267ms/epoch - 6ms/step Epoch 5/500 Epoch 00005: val_loss improved from 0.00965 to 0.00469, saving model to LSTM6.h5 43/43 - 0s - loss: 0.0071 - accuracy: 0.0000e+00 - val_loss: 0.0047 - val_accuracy: 0.0037 - lr: 0.0010 - 286ms/epoch - 7ms/step Epoch 6/500 Epoch 00006: val_loss did not improve from 0.00469 43/43 - 0s - loss: 0.0026 - accuracy: 0.0000e+00 - val_loss: 0.0049 - val_accuracy: 0.0037 - lr: 0.0010 - 289ms/epoch - 7ms/step Epoch 7/500 Epoch 00007: val_loss did not improve from 0.00469 43/43 - 0s - loss: 0.0030 - accuracy: 0.0000e+00 - val_loss: 0.0049 - val_accuracy: 0.0037 - lr: 0.0010 - 271ms/epoch - 6ms/step Epoch 8/500 Epoch 00008: val_loss did not improve from 0.00469 43/43 - 0s - loss: 0.0022 - accuracy: 0.0000e+00 - val_loss: 0.0055 - val_accuracy: 0.0037 - lr: 0.0010 - 278ms/epoch - 6ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.00469 43/43 - 0s - loss: 0.0023 - accuracy: 0.0000e+00 - val_loss: 0.0096 - val_accuracy: 0.0037 - lr: 0.0010 - 241ms/epoch - 6ms/step Epoch 10/500 Epoch 00010: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00010: val_loss did not improve from 0.00469 43/43 - 0s - loss: 0.0024 - accuracy: 0.0000e+00 - val_loss: 0.0119 - val_accuracy: 0.0037 - lr: 0.0010 - 235ms/epoch - 5ms/step Epoch 11/500 Epoch 00011: val_loss did not improve from 0.00469 43/43 - 0s - loss: 0.0025 - accuracy: 0.0000e+00 - val_loss: 0.0252 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 279ms/epoch - 6ms/step Epoch 12/500 Epoch 00012: val_loss did not improve from 0.00469 43/43 - 0s - loss: 0.0019 - accuracy: 0.0000e+00 - val_loss: 0.0176 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 275ms/epoch - 6ms/step Epoch 13/500 Epoch 00013: val_loss did not improve from 0.00469 43/43 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0160 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 235ms/epoch - 5ms/step Epoch 14/500 Epoch 00014: val_loss did not improve from 0.00469 43/43 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0151 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 250ms/epoch - 6ms/step Epoch 15/500 Epoch 00015: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00015: val_loss did not improve from 0.00469 43/43 - 0s - loss: 9.6727e-04 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 233ms/epoch - 5ms/step Epoch 16/500 Epoch 00016: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.9876e-04 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 236ms/epoch - 5ms/step Epoch 17/500 Epoch 00017: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.9638e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 234ms/epoch - 5ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.9429e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 239ms/epoch - 6ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.9234e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 283ms/epoch - 7ms/step Epoch 20/500 Epoch 00020: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00020: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.9045e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 255ms/epoch - 6ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.8859e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 240ms/epoch - 6ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.8677e-04 - accuracy: 0.0000e+00 - val_loss: 0.0145 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 242ms/epoch - 6ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.8496e-04 - accuracy: 0.0000e+00 - val_loss: 0.0144 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 254ms/epoch - 6ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.8318e-04 - accuracy: 0.0000e+00 - val_loss: 0.0144 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 252ms/epoch - 6ms/step Epoch 25/500 Epoch 00025: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.8142e-04 - accuracy: 0.0000e+00 - val_loss: 0.0144 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 276ms/epoch - 6ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.7968e-04 - accuracy: 0.0000e+00 - val_loss: 0.0144 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 268ms/epoch - 6ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.7796e-04 - accuracy: 0.0000e+00 - val_loss: 0.0144 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 254ms/epoch - 6ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.7627e-04 - accuracy: 0.0000e+00 - val_loss: 0.0144 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 234ms/epoch - 5ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.7460e-04 - accuracy: 0.0000e+00 - val_loss: 0.0144 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 246ms/epoch - 6ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.7295e-04 - accuracy: 0.0000e+00 - val_loss: 0.0143 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 280ms/epoch - 7ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.7132e-04 - accuracy: 0.0000e+00 - val_loss: 0.0143 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 263ms/epoch - 6ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.6972e-04 - accuracy: 0.0000e+00 - val_loss: 0.0143 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 278ms/epoch - 6ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.6814e-04 - accuracy: 0.0000e+00 - val_loss: 0.0143 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 241ms/epoch - 6ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.6658e-04 - accuracy: 0.0000e+00 - val_loss: 0.0143 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 251ms/epoch - 6ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.6504e-04 - accuracy: 0.0000e+00 - val_loss: 0.0143 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 251ms/epoch - 6ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.6352e-04 - accuracy: 0.0000e+00 - val_loss: 0.0143 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 268ms/epoch - 6ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.6202e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 273ms/epoch - 6ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.6055e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 286ms/epoch - 7ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.5909e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 274ms/epoch - 6ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.5765e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 252ms/epoch - 6ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.5624e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 271ms/epoch - 6ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.5484e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 230ms/epoch - 5ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.5346e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 242ms/epoch - 6ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.5209e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 283ms/epoch - 7ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.5074e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 277ms/epoch - 6ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.4941e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 278ms/epoch - 6ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.4809e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 277ms/epoch - 6ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.4679e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 273ms/epoch - 6ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.4550e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 255ms/epoch - 6ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.4422e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 235ms/epoch - 5ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.4296e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 248ms/epoch - 6ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.4170e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 233ms/epoch - 5ms/step Epoch 53/500 Epoch 00053: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.4046e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 249ms/epoch - 6ms/step Epoch 54/500 Epoch 00054: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.3922e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 259ms/epoch - 6ms/step Epoch 55/500 Epoch 00055: val_loss did not improve from 0.00469 43/43 - 0s - loss: 8.3800e-04 - accuracy: 0.0000e+00 - val_loss: 0.0143 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 266ms/epoch - 6ms/step Epoch 00055: early stopping
SMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 109.05489356902385
RMSE: 10.442935103170173
MAPE: 8.732625329283675
EMA
Prediction vs Close: 54.85% Accuracy
Prediction vs Prediction: 44.4% Accuracy
MSE: 73.04142380966745
RMSE: 8.546427546622475
MAPE: 7.099244401385842
WMA
Prediction vs Close: 53.36% Accuracy
Prediction vs Prediction: 44.78% Accuracy
MSE: 84.94978866341171
RMSE: 9.21682096296829
MAPE: 7.490547440692417
DEMA
Prediction vs Close: 52.24% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 151.61070955572364
RMSE: 12.313030072070955
MAPE: 11.085595013418024
KAMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 80.0014001509936
RMSE: 8.944350180476702
MAPE: 7.358601729961256
MIDPOINT
Prediction vs Close: 52.61% Accuracy
Prediction vs Prediction: 43.66% Accuracy
MSE: 61.18379984959283
RMSE: 7.822007405365507
MAPE: 6.441728946960992
T3
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 46.27% Accuracy
MSE: 110.56518298054853
RMSE: 10.514998001927937
MAPE: 8.473394546481362
TEMA
TEMA([input_arrays], [timeperiod=30])
Triple Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
9
Working on TEMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16762.799, Time=4.59 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14158.507, Time=2.83 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-16445.598, Time=8.85 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-16144.282, Time=11.24 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-15275.101, Time=9.09 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-15897.090, Time=13.07 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-16446.973, Time=9.72 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=-16567.628, Time=3.34 sec
/usr/local/lib/python3.7/dist-packages/statsmodels/tsa/statespace/sarimax.py:1906: RuntimeWarning: divide by zero encountered in reciprocal return np.roots(self.polynomial_reduced_ma)**-1
ARIMA(2,3,2)(0,0,0)[0] : AIC=-16523.926, Time=3.78 sec
ARIMA(1,3,1)(0,0,0)[0] intercept : AIC=-16696.008, Time=3.35 sec
Best model: ARIMA(1,3,1)(0,0,0)[0]
Total fit time: 69.865 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(1, 3, 1) Log Likelihood 8408.400
Date: Sun, 12 Dec 2021 AIC -16762.799
Time: 16:52:44 BIC -16636.147
Sample: 0 HQIC -16714.159
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -5.289e-07 0.001 -0.000 1.000 -0.002 0.002
x2 -5.288e-07 0.001 -0.001 0.999 -0.002 0.002
x3 -5.306e-07 0.001 -0.000 1.000 -0.002 0.002
x4 1.0000 0.000 2045.695 0.000 0.999 1.001
x5 -5.041e-07 0.000 -0.001 0.999 -0.001 0.001
x6 -9.879e-07 4.33e-05 -0.023 0.982 -8.58e-05 8.38e-05
x7 -5.185e-07 0.001 -0.001 0.999 -0.001 0.001
x8 0.0001 0.000 0.643 0.520 -0.000 0.001
x9 9.794e-08 0.001 0.000 1.000 -0.001 0.001
x10 0.0001 0.000 0.313 0.754 -0.001 0.001
x11 -0.0004 0.000 -2.284 0.022 -0.001 -6.06e-05
x12 0.0005 0.000 2.453 0.014 0.000 0.001
x13 -5.277e-07 0.000 -0.002 0.999 -0.001 0.001
x14 -1.566e-06 0.000 -0.005 0.996 -0.001 0.001
x15 -5.136e-07 9.86e-05 -0.005 0.996 -0.000 0.000
x16 -7.66e-07 0.000 -0.002 0.999 -0.001 0.001
x17 -5.146e-07 0.000 -0.003 0.998 -0.000 0.000
x18 -1.701e-07 0.001 -0.000 1.000 -0.001 0.001
x19 -5.77e-07 8.54e-05 -0.007 0.995 -0.000 0.000
x20 5.026e-07 0.001 0.001 0.999 -0.001 0.001
x21 -2.058e-06 0.000 -0.010 0.992 -0.000 0.000
x22 -1.098e-06 0.001 -0.001 0.999 -0.003 0.003
x23 -1.472e-06 0.001 -0.003 0.998 -0.001 0.001
x24 -8.255e-07 0.001 -0.001 0.999 -0.002 0.002
ar.L1 -0.2866 3.63e-05 -7897.273 0.000 -0.287 -0.287
ma.L1 -0.9124 1.46e-06 -6.25e+05 0.000 -0.912 -0.912
sigma2 9.98e-11 7.23e-11 1.380 0.168 -4.2e-11 2.42e-10
===================================================================================
Ljung-Box (L1) (Q): 83.51 Jarque-Bera (JB): 4742889.91
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: -5.71
Prob(H) (two-sided): 0.00 Kurtosis: 378.86
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 4.2e+22. Standard errors may be unstable.
ARIMA order: (1, 3, 1)
/usr/local/lib/python3.7/dist-packages/keras/optimizer_v2/adam.py:105: UserWarning: The `lr` argument is deprecated, use `learning_rate` instead. super(Adam, self).__init__(name, **kwargs)
Epoch 1/500 Epoch 00001: val_loss improved from inf to 0.05037, saving model to LSTM6.h5 90/90 - 5s - loss: 0.0961 - accuracy: 0.0000e+00 - val_loss: 0.0504 - val_accuracy: 0.0037 - lr: 0.0010 - 5s/epoch - 50ms/step Epoch 2/500 Epoch 00002: val_loss did not improve from 0.05037 90/90 - 0s - loss: 0.1160 - accuracy: 0.0000e+00 - val_loss: 0.2484 - val_accuracy: 0.0000e+00 - lr: 0.0010 - 482ms/epoch - 5ms/step Epoch 3/500 Epoch 00003: val_loss improved from 0.05037 to 0.01092, saving model to LSTM6.h5 90/90 - 1s - loss: 0.0670 - accuracy: 0.0000e+00 - val_loss: 0.0109 - val_accuracy: 0.0037 - lr: 0.0010 - 529ms/epoch - 6ms/step Epoch 4/500 Epoch 00004: val_loss did not improve from 0.01092 90/90 - 1s - loss: 0.0270 - accuracy: 0.0000e+00 - val_loss: 0.0156 - val_accuracy: 0.0037 - lr: 0.0010 - 503ms/epoch - 6ms/step Epoch 5/500 Epoch 00005: val_loss did not improve from 0.01092 90/90 - 1s - loss: 0.0116 - accuracy: 0.0000e+00 - val_loss: 0.0272 - val_accuracy: 0.0037 - lr: 0.0010 - 527ms/epoch - 6ms/step Epoch 6/500 Epoch 00006: val_loss did not improve from 0.01092 90/90 - 1s - loss: 0.0106 - accuracy: 0.0000e+00 - val_loss: 0.0153 - val_accuracy: 0.0037 - lr: 0.0010 - 512ms/epoch - 6ms/step Epoch 7/500 Epoch 00007: val_loss did not improve from 0.01092 90/90 - 1s - loss: 0.0075 - accuracy: 0.0000e+00 - val_loss: 0.0133 - val_accuracy: 0.0037 - lr: 0.0010 - 535ms/epoch - 6ms/step Epoch 8/500 Epoch 00008: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513. Epoch 00008: val_loss did not improve from 0.01092 90/90 - 1s - loss: 0.0119 - accuracy: 0.0000e+00 - val_loss: 0.0256 - val_accuracy: 0.0037 - lr: 0.0010 - 509ms/epoch - 6ms/step Epoch 9/500 Epoch 00009: val_loss did not improve from 0.01092 90/90 - 1s - loss: 0.0135 - accuracy: 0.0000e+00 - val_loss: 0.0167 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 574ms/epoch - 6ms/step Epoch 10/500 Epoch 00010: val_loss improved from 0.01092 to 0.01013, saving model to LSTM6.h5 90/90 - 1s - loss: 0.0100 - accuracy: 0.0000e+00 - val_loss: 0.0101 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 515ms/epoch - 6ms/step Epoch 11/500 Epoch 00011: val_loss improved from 0.01013 to 0.00804, saving model to LSTM6.h5 90/90 - 0s - loss: 0.0068 - accuracy: 0.0000e+00 - val_loss: 0.0080 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 497ms/epoch - 6ms/step Epoch 12/500 Epoch 00012: val_loss improved from 0.00804 to 0.00696, saving model to LSTM6.h5 90/90 - 1s - loss: 0.0053 - accuracy: 0.0000e+00 - val_loss: 0.0070 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 551ms/epoch - 6ms/step Epoch 13/500 Epoch 00013: val_loss improved from 0.00696 to 0.00632, saving model to LSTM6.h5 90/90 - 1s - loss: 0.0043 - accuracy: 0.0000e+00 - val_loss: 0.0063 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 512ms/epoch - 6ms/step Epoch 14/500 Epoch 00014: val_loss improved from 0.00632 to 0.00592, saving model to LSTM6.h5 90/90 - 1s - loss: 0.0036 - accuracy: 0.0000e+00 - val_loss: 0.0059 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 506ms/epoch - 6ms/step Epoch 15/500 Epoch 00015: val_loss improved from 0.00592 to 0.00572, saving model to LSTM6.h5 90/90 - 1s - loss: 0.0030 - accuracy: 0.0000e+00 - val_loss: 0.0057 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 562ms/epoch - 6ms/step Epoch 16/500 Epoch 00016: val_loss improved from 0.00572 to 0.00568, saving model to LSTM6.h5 90/90 - 1s - loss: 0.0025 - accuracy: 0.0000e+00 - val_loss: 0.0057 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 518ms/epoch - 6ms/step Epoch 17/500 Epoch 00017: val_loss did not improve from 0.00568 90/90 - 0s - loss: 0.0022 - accuracy: 0.0000e+00 - val_loss: 0.0058 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 482ms/epoch - 5ms/step Epoch 18/500 Epoch 00018: val_loss did not improve from 0.00568 90/90 - 0s - loss: 0.0019 - accuracy: 0.0000e+00 - val_loss: 0.0062 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 454ms/epoch - 5ms/step Epoch 19/500 Epoch 00019: val_loss did not improve from 0.00568 90/90 - 1s - loss: 0.0017 - accuracy: 0.0000e+00 - val_loss: 0.0067 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 502ms/epoch - 6ms/step Epoch 20/500 Epoch 00020: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05. Epoch 00020: val_loss did not improve from 0.00568 90/90 - 0s - loss: 0.0015 - accuracy: 0.0000e+00 - val_loss: 0.0074 - val_accuracy: 0.0037 - lr: 1.0000e-04 - 491ms/epoch - 5ms/step Epoch 21/500 Epoch 00021: val_loss did not improve from 0.00568 90/90 - 1s - loss: 0.0013 - accuracy: 0.0000e+00 - val_loss: 0.0079 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 502ms/epoch - 6ms/step Epoch 22/500 Epoch 00022: val_loss did not improve from 0.00568 90/90 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0084 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 486ms/epoch - 5ms/step Epoch 23/500 Epoch 00023: val_loss did not improve from 0.00568 90/90 - 1s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0086 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 532ms/epoch - 6ms/step Epoch 24/500 Epoch 00024: val_loss did not improve from 0.00568 90/90 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0089 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 477ms/epoch - 5ms/step Epoch 25/500 Epoch 00025: ReduceLROnPlateau reducing learning rate to 1e-05. Epoch 00025: val_loss did not improve from 0.00568 90/90 - 1s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0091 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 608ms/epoch - 7ms/step Epoch 26/500 Epoch 00026: val_loss did not improve from 0.00568 90/90 - 0s - loss: 0.0012 - accuracy: 0.0000e+00 - val_loss: 0.0093 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 459ms/epoch - 5ms/step Epoch 27/500 Epoch 00027: val_loss did not improve from 0.00568 90/90 - 1s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0095 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 562ms/epoch - 6ms/step Epoch 28/500 Epoch 00028: val_loss did not improve from 0.00568 90/90 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0096 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 470ms/epoch - 5ms/step Epoch 29/500 Epoch 00029: val_loss did not improve from 0.00568 90/90 - 1s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0098 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 533ms/epoch - 6ms/step Epoch 30/500 Epoch 00030: val_loss did not improve from 0.00568 90/90 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0100 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 480ms/epoch - 5ms/step Epoch 31/500 Epoch 00031: val_loss did not improve from 0.00568 90/90 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0102 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 492ms/epoch - 5ms/step Epoch 32/500 Epoch 00032: val_loss did not improve from 0.00568 90/90 - 1s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0104 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 558ms/epoch - 6ms/step Epoch 33/500 Epoch 00033: val_loss did not improve from 0.00568 90/90 - 1s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0107 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 550ms/epoch - 6ms/step Epoch 34/500 Epoch 00034: val_loss did not improve from 0.00568 90/90 - 0s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0109 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 481ms/epoch - 5ms/step Epoch 35/500 Epoch 00035: val_loss did not improve from 0.00568 90/90 - 1s - loss: 0.0011 - accuracy: 0.0000e+00 - val_loss: 0.0111 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 504ms/epoch - 6ms/step Epoch 36/500 Epoch 00036: val_loss did not improve from 0.00568 90/90 - 1s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0114 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 557ms/epoch - 6ms/step Epoch 37/500 Epoch 00037: val_loss did not improve from 0.00568 90/90 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0116 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 471ms/epoch - 5ms/step Epoch 38/500 Epoch 00038: val_loss did not improve from 0.00568 90/90 - 0s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0119 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 483ms/epoch - 5ms/step Epoch 39/500 Epoch 00039: val_loss did not improve from 0.00568 90/90 - 1s - loss: 0.0010 - accuracy: 0.0000e+00 - val_loss: 0.0121 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 506ms/epoch - 6ms/step Epoch 40/500 Epoch 00040: val_loss did not improve from 0.00568 90/90 - 0s - loss: 9.9794e-04 - accuracy: 0.0000e+00 - val_loss: 0.0124 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 478ms/epoch - 5ms/step Epoch 41/500 Epoch 00041: val_loss did not improve from 0.00568 90/90 - 1s - loss: 9.8785e-04 - accuracy: 0.0000e+00 - val_loss: 0.0127 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 560ms/epoch - 6ms/step Epoch 42/500 Epoch 00042: val_loss did not improve from 0.00568 90/90 - 1s - loss: 9.7796e-04 - accuracy: 0.0000e+00 - val_loss: 0.0130 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 503ms/epoch - 6ms/step Epoch 43/500 Epoch 00043: val_loss did not improve from 0.00568 90/90 - 0s - loss: 9.6828e-04 - accuracy: 0.0000e+00 - val_loss: 0.0133 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 472ms/epoch - 5ms/step Epoch 44/500 Epoch 00044: val_loss did not improve from 0.00568 90/90 - 1s - loss: 9.5883e-04 - accuracy: 0.0000e+00 - val_loss: 0.0136 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 560ms/epoch - 6ms/step Epoch 45/500 Epoch 00045: val_loss did not improve from 0.00568 90/90 - 0s - loss: 9.4960e-04 - accuracy: 0.0000e+00 - val_loss: 0.0139 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 492ms/epoch - 5ms/step Epoch 46/500 Epoch 00046: val_loss did not improve from 0.00568 90/90 - 1s - loss: 9.4060e-04 - accuracy: 0.0000e+00 - val_loss: 0.0142 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 509ms/epoch - 6ms/step Epoch 47/500 Epoch 00047: val_loss did not improve from 0.00568 90/90 - 0s - loss: 9.3183e-04 - accuracy: 0.0000e+00 - val_loss: 0.0146 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 454ms/epoch - 5ms/step Epoch 48/500 Epoch 00048: val_loss did not improve from 0.00568 90/90 - 1s - loss: 9.2329e-04 - accuracy: 0.0000e+00 - val_loss: 0.0149 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 561ms/epoch - 6ms/step Epoch 49/500 Epoch 00049: val_loss did not improve from 0.00568 90/90 - 1s - loss: 9.1498e-04 - accuracy: 0.0000e+00 - val_loss: 0.0153 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 528ms/epoch - 6ms/step Epoch 50/500 Epoch 00050: val_loss did not improve from 0.00568 90/90 - 1s - loss: 9.0690e-04 - accuracy: 0.0000e+00 - val_loss: 0.0156 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 503ms/epoch - 6ms/step Epoch 51/500 Epoch 00051: val_loss did not improve from 0.00568 90/90 - 0s - loss: 8.9904e-04 - accuracy: 0.0000e+00 - val_loss: 0.0160 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 462ms/epoch - 5ms/step Epoch 52/500 Epoch 00052: val_loss did not improve from 0.00568 90/90 - 0s - loss: 8.9141e-04 - accuracy: 0.0000e+00 - val_loss: 0.0164 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 484ms/epoch - 5ms/step Epoch 53/500 Epoch 00053: val_loss did not improve from 0.00568 90/90 - 0s - loss: 8.8398e-04 - accuracy: 0.0000e+00 - val_loss: 0.0168 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 476ms/epoch - 5ms/step Epoch 54/500 Epoch 00054: val_loss did not improve from 0.00568 90/90 - 1s - loss: 8.7677e-04 - accuracy: 0.0000e+00 - val_loss: 0.0172 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 564ms/epoch - 6ms/step Epoch 55/500 Epoch 00055: val_loss did not improve from 0.00568 90/90 - 0s - loss: 8.6976e-04 - accuracy: 0.0000e+00 - val_loss: 0.0176 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 480ms/epoch - 5ms/step Epoch 56/500 Epoch 00056: val_loss did not improve from 0.00568 90/90 - 0s - loss: 8.6294e-04 - accuracy: 0.0000e+00 - val_loss: 0.0180 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 497ms/epoch - 6ms/step Epoch 57/500 Epoch 00057: val_loss did not improve from 0.00568 90/90 - 0s - loss: 8.5630e-04 - accuracy: 0.0000e+00 - val_loss: 0.0184 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 458ms/epoch - 5ms/step Epoch 58/500 Epoch 00058: val_loss did not improve from 0.00568 90/90 - 0s - loss: 8.4984e-04 - accuracy: 0.0000e+00 - val_loss: 0.0188 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 493ms/epoch - 5ms/step Epoch 59/500 Epoch 00059: val_loss did not improve from 0.00568 90/90 - 0s - loss: 8.4355e-04 - accuracy: 0.0000e+00 - val_loss: 0.0193 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 475ms/epoch - 5ms/step Epoch 60/500 Epoch 00060: val_loss did not improve from 0.00568 90/90 - 0s - loss: 8.3742e-04 - accuracy: 0.0000e+00 - val_loss: 0.0197 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 475ms/epoch - 5ms/step Epoch 61/500 Epoch 00061: val_loss did not improve from 0.00568 90/90 - 0s - loss: 8.3145e-04 - accuracy: 0.0000e+00 - val_loss: 0.0202 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 459ms/epoch - 5ms/step Epoch 62/500 Epoch 00062: val_loss did not improve from 0.00568 90/90 - 0s - loss: 8.2562e-04 - accuracy: 0.0000e+00 - val_loss: 0.0206 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 496ms/epoch - 6ms/step Epoch 63/500 Epoch 00063: val_loss did not improve from 0.00568 90/90 - 0s - loss: 8.1992e-04 - accuracy: 0.0000e+00 - val_loss: 0.0211 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 479ms/epoch - 5ms/step Epoch 64/500 Epoch 00064: val_loss did not improve from 0.00568 90/90 - 1s - loss: 8.1436e-04 - accuracy: 0.0000e+00 - val_loss: 0.0215 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 587ms/epoch - 7ms/step Epoch 65/500 Epoch 00065: val_loss did not improve from 0.00568 90/90 - 0s - loss: 8.0893e-04 - accuracy: 0.0000e+00 - val_loss: 0.0220 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 478ms/epoch - 5ms/step Epoch 66/500 Epoch 00066: val_loss did not improve from 0.00568 90/90 - 1s - loss: 8.0362e-04 - accuracy: 0.0000e+00 - val_loss: 0.0224 - val_accuracy: 0.0037 - lr: 1.0000e-05 - 540ms/epoch - 6ms/step Epoch 00066: early stopping
SMA Prediction vs Close: 52.99% Accuracy Prediction vs Prediction: 46.64% Accuracy MSE: 109.05489356902385 RMSE: 10.442935103170173 MAPE: 8.732625329283675 EMA Prediction vs Close: 54.85% Accuracy Prediction vs Prediction: 44.4% Accuracy MSE: 73.04142380966745 RMSE: 8.546427546622475 MAPE: 7.099244401385842 WMA Prediction vs Close: 53.36% Accuracy Prediction vs Prediction: 44.78% Accuracy MSE: 84.94978866341171 RMSE: 9.21682096296829 MAPE: 7.490547440692417 DEMA Prediction vs Close: 52.24% Accuracy Prediction vs Prediction: 45.9% Accuracy MSE: 151.61070955572364 RMSE: 12.313030072070955 MAPE: 11.085595013418024 KAMA Prediction vs Close: 52.99% Accuracy Prediction vs Prediction: 46.64% Accuracy MSE: 80.0014001509936 RMSE: 8.944350180476702 MAPE: 7.358601729961256 MIDPOINT Prediction vs Close: 52.61% Accuracy Prediction vs Prediction: 43.66% Accuracy MSE: 61.18379984959283 RMSE: 7.822007405365507 MAPE: 6.441728946960992 T3 Prediction vs Close: 54.1% Accuracy Prediction vs Prediction: 46.27% Accuracy MSE: 110.56518298054853 RMSE: 10.514998001927937 MAPE: 8.473394546481362 TEMA Prediction vs Close: 50.37% Accuracy Prediction vs Prediction: 49.63% Accuracy MSE: 69.56753550271695 RMSE: 8.340715527022663 MAPE: 7.185876850367952 Runtime: mins: 54.85666802865001
Architecture Used¶
In [128]:
from google.colab import files
import cv2
uploaded = files.upload()
Saving Experiment6.png to Experiment6 (2).png
In [130]:
img = cv2.imread('Experiment6.png')
plt.figure(figsize=(20,10))
plt.axis("off")
plt.title('LSTM Architecture '+imgfile,fontsize=18)
plt.imshow(img)
Out[130]:
<matplotlib.image.AxesImage at 0x7f7662efeb90>
Model Plots¶
In [106]:
with open('simulation6_data.json') as json_file:
simulation6 = json.load(json_file)
fileimg = 'Experiment6'
In [107]:
for i in range(len(list(simulation6.keys()))):
SIM = list(simulation6.keys())[i]
plot_train(simulation6,SIM)
plot_test(simulation6,SIM)
----- Train RMSE for SMA ----- 8.884485327179771 ----- Train_MSE_LSTM for SMA ----- 78.93407952887263 ----- Train MAE LSTM for SMA ----- 7.7230452013589295
----- Test RMSE for SMA----- 10.442935103170173 ----- Test_MSE_LSTM for SMA----- 109.05489356902385 ----- Test_MAE_LSTM for SMA----- 8.732625329283675
----- Train RMSE for EMA ----- 10.171650757359036 ----- Train_MSE_LSTM for EMA ----- 103.46247912968265 ----- Train MAE LSTM for EMA ----- 8.997792035349331
----- Test RMSE for EMA----- 8.546427546622475 ----- Test_MSE_LSTM for EMA----- 73.04142380966745 ----- Test_MAE_LSTM for EMA----- 7.099244401385842
----- Train RMSE for WMA ----- 10.46606060205442 ----- Train_MSE_LSTM for WMA ----- 109.53842452587571 ----- Train MAE LSTM for WMA ----- 9.32499706302078
----- Test RMSE for WMA----- 9.21682096296829 ----- Test_MSE_LSTM for WMA----- 84.94978866341171 ----- Test_MAE_LSTM for WMA----- 7.490547440692417
----- Train RMSE for DEMA ----- 12.135400340426562 ----- Train_MSE_LSTM for DEMA ----- 147.26794142242514 ----- Train MAE LSTM for DEMA ----- 10.923669650609977
----- Test RMSE for DEMA----- 12.313030072070955 ----- Test_MSE_LSTM for DEMA----- 151.61070955572364 ----- Test_MAE_LSTM for DEMA----- 11.085595013418024
----- Train RMSE for KAMA ----- 10.536710095405018 ----- Train_MSE_LSTM for KAMA ----- 111.02225963461002 ----- Train MAE LSTM for KAMA ----- 9.489313762266152
----- Test RMSE for KAMA----- 8.944350180476702 ----- Test_MSE_LSTM for KAMA----- 80.0014001509936 ----- Test_MAE_LSTM for KAMA----- 7.358601729961256
----- Train RMSE for MIDPOINT ----- 9.483224784900598 ----- Train_MSE_LSTM for MIDPOINT ----- 89.93155232095299 ----- Train MAE LSTM for MIDPOINT ----- 8.402424233592125
----- Test RMSE for MIDPOINT----- 7.822007405365507 ----- Test_MSE_LSTM for MIDPOINT----- 61.18379984959283 ----- Test_MAE_LSTM for MIDPOINT----- 6.441728946960992
----- Train RMSE for T3 ----- 12.056010756671938 ----- Train_MSE_LSTM for T3 ----- 145.3473953649895 ----- Train MAE LSTM for T3 ----- 10.855993463265904
----- Test RMSE for T3----- 10.514998001927937 ----- Test_MSE_LSTM for T3----- 110.56518298054853 ----- Test_MAE_LSTM for T3----- 8.473394546481362
----- Train RMSE for TEMA ----- 7.432705233614638 ----- Train_MSE_LSTM for TEMA ----- 55.24510708980243 ----- Train MAE LSTM for TEMA ----- 5.131892957309109
----- Test RMSE for TEMA----- 8.340715527022663 ----- Test_MSE_LSTM for TEMA----- 69.56753550271695 ----- Test_MAE_LSTM for TEMA----- 7.185876850367952
Arima w Exogenous Variable Multistep MutiVariate LSTM Hybrid Model Experiment 7¶
In [131]:
def get_arima_exog(dataframe,original_data, train_len, test_len):
# prepare train and test data for exogenous vr
X_value = pd.DataFrame(low_vol.iloc[:, :])
y_value = pd.DataFrame(low_vol.iloc[:, 3])
X_scaler = MinMaxScaler(feature_range=(-1, 1))
y_scaler = MinMaxScaler(feature_range=(-1, 1))
X_scaler.fit(X_value)
y_scaler.fit(y_value)
X_scale_dataset = X_scaler.fit_transform(X_value)
y_scale_dataset = y_scaler.fit_transform(y_value)
# Get data and check shape
# X, y, yc = get_X_y(X_scale_dataset, y_scale_dataset)#X will be of shape 224 X 3 X 21 (each 3 X 21 array will be 3 days' worth of data). yc will have the corresponding closing price value
# pdb.set_trace()
X_train, X_test, = split_train_test(X_scale_dataset)
y_train, y_test, = split_train_test(y_scale_dataset)
yc_train,yc_test = split_train_test(low_vol_data)
yc = yc_test.values.tolist()
y_train_list = y_train.flatten().tolist()
y_test_list = y_test.flatten().tolist()
# yc_train, yc_test, = split_train_test(original_data)
index_train, index_test, = predict_index(dataset_final, X_train, n_steps_in, n_steps_out)
# Initialize model
model = auto_arima(y_train_list,exogenous = X_train,trace=True, error_action='ignore', start_p=1,start_q=1,max_p=3,max_q=3,d=3,
suppress_warnings=True,stepwise=True,seasonal=True)
# Determine model parameters
print(model.summary())
model.fit(y_train_list,maxiter=200)
order = model.get_params()['order']
print('ARIMA order:', order, '\n')
# Genereate predictions
prediction = []
for i in range(len(y_test_list)):
model = pmdarima.ARIMA(order=order)
model.fit(y_train_list)
# print('working on', i+1, 'of', len(y_test), '-- ' + str(int(100 * (i + 1) / len(y_test))) + '% complete')
prediction.append(model.predict()[0])
y_train_list.append(y_test_list[i])
predictionte = y_scaler.inverse_transform(np.array(prediction).reshape(-1,1))
y_test_ = y_scaler.inverse_transform(np.array(y_test_list).reshape(-1,1))
# Generate error data
mse = mean_squared_error(yc_test, predictionte)
rmse = mse ** 0.5
mae = mean_absolute_error(y_test_ , predictionte )
return yc,predictionte.flatten().tolist(), mse, rmse, mae
In [132]:
def get_lstm(data,original_data, train_len, test_len,img_file,ma ,lstm_len=3):
# prepare train and test data
X_value = pd.DataFrame(data.iloc[:, :])
y_value = pd.DataFrame(data.iloc[:, 3])
X_scaler = MinMaxScaler(feature_range=(-1, 1))
y_scaler = MinMaxScaler(feature_range=(-1, 1))
X_scaler.fit(X_value)
y_scaler.fit(y_value)
# Get data and check shape
X, y, yc = get_X_y(X_scale_dataset, y_scale_dataset)#X will be of shape 224 X 3 X 21 (each 3 X 21 array will be 3 days' worth of data). yc will have the corresponding closing price value
# pdb.set_trace()
X_train, X_test, = split_train_test(X)
y_train, y_test, = split_train_test(y)
# yc_train, yc_test, = split_train_test(original_data)
index_train, index_test, = predict_index(dataset_final, X_train, n_steps_in, n_steps_out)
det =20
input_dim = X_train.shape[1]#3
feature_size = X_train.shape[2]#24
output_dim = y_train.shape[1]#1
# Option 1
# Set up & fit LSTM RNN
# model = Sequential()
# model.add(LSTM(256, activation='relu', kernel_initializer='he_normal', input_shape=(input_dim, feature_size)))
# model.add(Dense(units=64,activation='relu'))
# model.add(Dropout(0.5))
# model.add(Dense(units=output_dim))
# model.compile(optimizer=Adam(learning_rate = 0.001), loss='mse')
# ## Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# # # option 2
# model = Sequential()
# model.add(Bidirectional(LSTM(units= 128), input_shape=(input_dim, feature_size)))
# model.add(Dense(64))
# model.add(Dense(units=output_dim))
# model.compile(optimizer=Adam(lr = 0.001), loss='mean_squared_error', metrics=['accuracy'])
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM7.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma+' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# Option 3
# define custom activation
#
class Double_Tanh(Activation):
def __init__(self, activation, **kwargs):
super(Double_Tanh, self).__init__(activation, **kwargs)
self.__name__ = 'double_tanh'
def double_tanh(x):
return (K.tanh(x) * 2)
get_custom_objects().update({'double_tanh':Double_Tanh(double_tanh)})
# Model Generation
model = Sequential()
#check https://machinelearningmastery.com/use-weight-regularization-lstm-networks-time-series-forecasting/
model.add(LSTM(25, input_shape=(input_dim, feature_size), dropout=0.2, kernel_regularizer=l1_l2(0.00,0.00), bias_regularizer=l1_l2(0.00,0.00)))
model.add(Dense(1))
model.add(Activation(double_tanh))
model.compile(loss='mean_squared_error', optimizer='adam', metrics=['mse', 'mae'])
# Common code
callbacks = [
EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
ModelCheckpoint('LSTM7.h5', verbose=1, save_best_only=True, save_weights_only=True)]
fname1 = img_file+'.png'
tensorflow.keras.utils.plot_model(
model, to_file=fname1, show_shapes=True, show_dtype=False,
show_layer_names=True, expand_nested=False, dpi=96,
layer_range=None, show_layer_activations=False
)
history = model.fit(X_train, y_train, epochs=500, batch_size=int( optimized_period[ma]), verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# plot loss
fname2 = img_file+'-'+ma
plt.title(img_file+'-'+ma+' Loss')
plt.xlabel("Epochs")
plt.ylabel("Loss")
pyplot.plot(history.history['loss'], label='train')
pyplot.plot(history.history['val_loss'], label='validation')
pyplot.legend()
pyplot.savefig(fname2+'.png',dpi='figure')
pyplot.show()
# Option 4
# Set up & fit LSTM RNN
# model = Sequential()
# model.add(LSTM(units=lstm_len, return_sequences=True, input_shape=(x_train.shape[1], 1)))
# model.add(LSTM(units=int(lstm_len/2)))
# model.add(Dense(1, activation='sigmoid'))
# model.compile(loss='mean_squared_error', optimizer='adam')
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# Generate predictions
predictiontr = model.predict(X_train, verbose=0)
predictiontr = y_scaler.inverse_transform(predictiontr).tolist()
outputtr = []
for i in range(len(predictiontr)):
outputtr.extend(predictiontr[i])
predictiontr = outputtr
# Generate error data
## replace with yc , xtest generated by new multistep method
mse_tr = mean_squared_error(y_train, predictiontr)
rmse_tr = mse_tr ** 0.5
# mape = mean_absolute_percentage_error(X_test, pd.Series(predictiontr))
mae_tr = mean_absolute_error(y_train, pd.Series(predictiontr))
# Original_tr = pd.Series(yc_train)
Original_tr = y_scaler.inverse_transform(y_train).flatten().tolist()
predictionte = model.predict(X_test, verbose=0)
predictionte = (y_scaler.inverse_transform(predictionte)-det).tolist()
outputte = []
for i in range(len(predictionte)):
outputte.extend(predictionte[i])
predictionte = outputte
# Generate error data
mse_te = mean_squared_error(y_test, predictionte)
rmse_te = mse_te ** 0.5
# mape = mean_absolute_percentage_error(X_test, pd.Series(predictionte))
mae_te = mean_absolute_error(y_test, pd.Series(predictionte))
# Original_te = pd.Series(yc_test)
Original_te = y_scaler.inverse_transform(y_test).flatten().tolist()
return Original_tr, predictiontr, mse_tr, rmse_tr,mae_tr,Original_te,predictionte, mse_te,rmse_te,mae_te
In [133]:
if __name__ == '__main__':
start_time = timeit.default_timer()
simulation7 = {}
imgfile = 'Experiment7'
for ma in optimized_period:
print(ma)
print(functions[ma])
print ( int( optimized_period[ma]))
# if ma == 'SMA':
low_vol = df.apply(lambda c: functions[ma](c, timeperiod = int( optimized_period[ma])))
low_vol = low_vol.fillna(0)
low_vol_data = df['close']
high_vol = pd.DataFrame()
df2 = df.copy()
for i in df2.columns:
if i in low_vol.columns:
high_vol[i] = df2[i].subtract(low_vol[i], fill_value=0)
high_vol_data = df['close']
## *****************************************************
# Generate ARIMA and LSTM predictions
print('\nWorking on ' + ma + ' predictions')
try:
print('parameters used : ', train_len, test_len)
low_vol_Original, low_vol_prediction, low_vol_mse, low_vol_rmse,low_vol_mae = get_arima_exog(low_vol,low_vol_data, train_len, test_len)
except:
print('ARIMA error, skipping to next MA type')
continue
Original_tr, predictiontr, mse_tr, rmse_tr,mae_tr,high_vol_Original, high_vol_prediction, high_vol_mse, high_vol_rmse,high_vol_mae, = get_lstm(high_vol,high_vol_data, train_len, test_len,imgfile,ma)
final_prediction_tr = df['close'].head(train_len).values + pd.Series(predictiontr) # ignoring first 3 steps
mse_ftr = mean_squared_error(df['close'].head(train_len).values,final_prediction_tr.values)
rmse_ftr = mse_ftr ** 0.5
mape_ftr = mean_absolute_percentage_error(df['close'].head(train_len).reset_index(drop=True), final_prediction_tr)
mae_ftr = mean_absolute_error(df['close'].head(train_len).reset_index(drop=True), final_prediction_tr)
final_prediction = pd.Series(low_vol_prediction[3:]) + pd.Series(high_vol_prediction)
mse = mean_squared_error(df['close'].tail(test_len).values,final_prediction.values)
rmse = mse ** 0.5
mape = mean_absolute_percentage_error(df['close'].tail(test_len).reset_index(drop=True), final_prediction)
mae = mean_absolute_error(df['close'].tail(test_len).reset_index(drop=True), final_prediction)
# Generate prediction accuracy
actual = df['close'].tail(test_len).values
result_1 = []
result_2 = []
for i in range(1, len(final_prediction)):
# Compare prediction to previous close price
if final_prediction[i] > actual[i-1] and actual[i] > actual[i-1]:
result_1.append(1)
elif final_prediction[i] < actual[i-1] and actual[i] < actual[i-1]:
result_1.append(1)
else:
result_1.append(0)
# Compare prediction to previous prediction
if final_prediction[i] > final_prediction[i-1] and actual[i] > actual[i-1]:
result_2.append(1)
elif final_prediction[i] < final_prediction[i-1] and actual[i] < actual[i-1]:
result_2.append(1)
else:
result_2.append(0)
accuracy_1 = np.mean(result_1)
accuracy_2 = np.mean(result_2)
simulation7[ma] = {'low_vol': {'original':list(low_vol_Original), 'prediction': list(low_vol_prediction) , 'mse': low_vol_mse,
'rmse': low_vol_rmse, 'mae' : low_vol_mae},
'high_vol': {'original':list(high_vol_Original),'prediction': list(high_vol_prediction), 'mse': high_vol_mse,
'rmse': high_vol_rmse, 'mae' : high_vol_mae},
'final_tr': {'original':df['close'].head(train_len).tolist(),'prediction': final_prediction_tr.values.tolist(), 'mse': mse_ftr,
'rmse': rmse_ftr, 'mae' : mae_ftr},
'final': {'original': df['close'].tail(test_len).tolist(), 'prediction': final_prediction.values.tolist(), 'mse': mse,
'rmse': rmse, 'mae': mae },
'accuracy': {'prediction vs close': accuracy_1, 'prediction vs prediction': accuracy_2}}
# save simulation data here as checkpoint
with open('simulation7_data.json', 'w') as fp:
json.dump(simulation7, fp)
for ma in simulation7.keys():
print('\n' + ma)
print('Prediction vs Close:\t\t' + str(round(100*simulation7[ma]['accuracy']['prediction vs close'], 2))
+ '% Accuracy')
print('Prediction vs Prediction:\t' + str(round(100*simulation7[ma]['accuracy']['prediction vs prediction'], 2))
+ '% Accuracy')
print('MSE:\t', simulation7[ma]['final']['mse'],
'\nRMSE:\t', simulation7[ma]['final']['rmse'],
'\nMAPE:\t', simulation7[ma]['final']['mae'])#,
# '\nMAPE:\t', simulation[ma]['final']['mape'])
# else:
# break
elapsed = timeit.default_timer() - start_time
print('Runtime: mins:',elapsed/60)
SMA
SMA([input_arrays], [timeperiod=30])
Simple Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
17
Working on SMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.786, Time=3.39 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.592, Time=4.80 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-15578.394, Time=8.54 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.592, Time=7.06 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16966.361, Time=9.28 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-16121.635, Time=10.17 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17214.069, Time=13.33 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-14562.592, Time=9.04 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-14572.319, Time=10.11 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-14403.474, Time=43.98 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 119.725 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8634.035
Date: Sun, 12 Dec 2021 AIC -17214.069
Time: 17:05:38 BIC -17087.416
Sample: 0 HQIC -17165.429
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -4.257e-09 9.55e-06 -0.000 1.000 -1.87e-05 1.87e-05
x2 -4.256e-09 9.56e-06 -0.000 1.000 -1.87e-05 1.87e-05
x3 -4.313e-09 9.62e-06 -0.000 1.000 -1.89e-05 1.88e-05
x4 1.0000 9.61e-06 1.04e+05 0.000 1.000 1.000
x5 -3.891e-09 9.14e-06 -0.000 1.000 -1.79e-05 1.79e-05
x6 -1.122e-08 1.03e-05 -0.001 0.999 -2.03e-05 2.03e-05
x7 -4.223e-09 9.54e-06 -0.000 1.000 -1.87e-05 1.87e-05
x8 -4.234e-09 9.55e-06 -0.000 1.000 -1.87e-05 1.87e-05
x9 -1.626e-10 6.54e-07 -0.000 1.000 -1.28e-06 1.28e-06
x10 -6.831e-10 2.91e-06 -0.000 1.000 -5.7e-06 5.7e-06
x11 -4.115e-09 9.41e-06 -0.000 1.000 -1.84e-05 1.84e-05
x12 -4.303e-09 9.62e-06 -0.000 1.000 -1.89e-05 1.88e-05
x13 -4.288e-09 9.6e-06 -0.000 1.000 -1.88e-05 1.88e-05
x14 -3.749e-08 2.81e-05 -0.001 0.999 -5.51e-05 5.5e-05
x15 -5.032e-09 1.04e-05 -0.000 1.000 -2.04e-05 2.03e-05
x16 -3.685e-09 9e-06 -0.000 1.000 -1.76e-05 1.76e-05
x17 -3.286e-09 8.45e-06 -0.000 1.000 -1.66e-05 1.66e-05
x18 -1.22e-08 1.59e-05 -0.001 0.999 -3.11e-05 3.11e-05
x19 -5.685e-09 1.1e-05 -0.001 1.000 -2.16e-05 2.16e-05
x20 -1.42e-08 1.69e-05 -0.001 0.999 -3.32e-05 3.32e-05
x21 -5.194e-08 3.31e-05 -0.002 0.999 -6.49e-05 6.48e-05
x22 -2.548e-08 2.31e-05 -0.001 0.999 -4.53e-05 4.52e-05
x23 -3.534e-08 2.73e-05 -0.001 0.999 -5.35e-05 5.34e-05
x24 -1.566e-08 1.8e-05 -0.001 0.999 -3.53e-05 3.53e-05
ma.L1 -1.3899 4.98e-09 -2.79e+08 0.000 -1.390 -1.390
ma.L2 0.4032 4.98e-09 8.09e+07 0.000 0.403 0.403
sigma2 7.635e-11 6.92e-11 1.103 0.270 -5.93e-11 2.12e-10
===================================================================================
Ljung-Box (L1) (Q): 68.48 Jarque-Bera (JB): 5579791.06
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 10.12
Prob(H) (two-sided): 0.00 Kurtosis: 410.36
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.69e+24. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 1.09957, saving model to LSTM7.h5
48/48 - 2s - loss: 0.1505 - mse: 0.1505 - mae: 0.3093 - val_loss: 1.0996 - val_mse: 1.0996 - val_mae: 1.0076 - lr: 0.0010 - 2s/epoch - 51ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 1.09957 to 0.56928, saving model to LSTM7.h5
48/48 - 0s - loss: 0.0396 - mse: 0.0396 - mae: 0.1611 - val_loss: 0.5693 - val_mse: 0.5693 - val_mae: 0.7083 - lr: 0.0010 - 241ms/epoch - 5ms/step
Epoch 3/500
Epoch 00003: val_loss improved from 0.56928 to 0.47229, saving model to LSTM7.h5
48/48 - 0s - loss: 0.0254 - mse: 0.0254 - mae: 0.1271 - val_loss: 0.4723 - val_mse: 0.4723 - val_mae: 0.6385 - lr: 0.0010 - 245ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0169 - mse: 0.0169 - mae: 0.1027 - val_loss: 0.5115 - val_mse: 0.5115 - val_mae: 0.6654 - lr: 0.0010 - 198ms/epoch - 4ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0113 - mse: 0.0113 - mae: 0.0845 - val_loss: 0.5410 - val_mse: 0.5410 - val_mae: 0.6857 - lr: 0.0010 - 234ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0101 - mse: 0.0101 - mae: 0.0792 - val_loss: 0.5219 - val_mse: 0.5219 - val_mae: 0.6726 - lr: 0.0010 - 231ms/epoch - 5ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0081 - mse: 0.0081 - mae: 0.0689 - val_loss: 0.5238 - val_mse: 0.5238 - val_mae: 0.6751 - lr: 0.0010 - 233ms/epoch - 5ms/step
Epoch 8/500
Epoch 00008: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00008: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0634 - val_loss: 0.5717 - val_mse: 0.5717 - val_mae: 0.7081 - lr: 0.0010 - 231ms/epoch - 5ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0584 - val_loss: 0.5696 - val_mse: 0.5696 - val_mae: 0.7067 - lr: 1.0000e-04 - 226ms/epoch - 5ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0063 - mse: 0.0063 - mae: 0.0614 - val_loss: 0.5560 - val_mse: 0.5560 - val_mae: 0.6979 - lr: 1.0000e-04 - 234ms/epoch - 5ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0602 - val_loss: 0.5515 - val_mse: 0.5515 - val_mae: 0.6949 - lr: 1.0000e-04 - 213ms/epoch - 4ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0592 - val_loss: 0.5506 - val_mse: 0.5506 - val_mae: 0.6945 - lr: 1.0000e-04 - 231ms/epoch - 5ms/step
Epoch 13/500
Epoch 00013: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00013: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0580 - val_loss: 0.5461 - val_mse: 0.5461 - val_mae: 0.6916 - lr: 1.0000e-04 - 197ms/epoch - 4ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0571 - val_loss: 0.5456 - val_mse: 0.5456 - val_mae: 0.6912 - lr: 1.0000e-05 - 204ms/epoch - 4ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0052 - mse: 0.0052 - mae: 0.0546 - val_loss: 0.5454 - val_mse: 0.5454 - val_mae: 0.6911 - lr: 1.0000e-05 - 212ms/epoch - 4ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0054 - mse: 0.0054 - mae: 0.0567 - val_loss: 0.5446 - val_mse: 0.5446 - val_mae: 0.6906 - lr: 1.0000e-05 - 210ms/epoch - 4ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0561 - val_loss: 0.5436 - val_mse: 0.5436 - val_mae: 0.6899 - lr: 1.0000e-05 - 204ms/epoch - 4ms/step
Epoch 18/500
Epoch 00018: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00018: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0593 - val_loss: 0.5428 - val_mse: 0.5428 - val_mae: 0.6894 - lr: 1.0000e-05 - 227ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0058 - mse: 0.0058 - mae: 0.0596 - val_loss: 0.5426 - val_mse: 0.5426 - val_mae: 0.6892 - lr: 1.0000e-05 - 216ms/epoch - 5ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0575 - val_loss: 0.5421 - val_mse: 0.5421 - val_mae: 0.6889 - lr: 1.0000e-05 - 202ms/epoch - 4ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0591 - val_loss: 0.5408 - val_mse: 0.5408 - val_mae: 0.6881 - lr: 1.0000e-05 - 217ms/epoch - 5ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0063 - mse: 0.0063 - mae: 0.0607 - val_loss: 0.5397 - val_mse: 0.5397 - val_mae: 0.6873 - lr: 1.0000e-05 - 189ms/epoch - 4ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0580 - val_loss: 0.5374 - val_mse: 0.5374 - val_mae: 0.6858 - lr: 1.0000e-05 - 233ms/epoch - 5ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0608 - val_loss: 0.5368 - val_mse: 0.5368 - val_mae: 0.6854 - lr: 1.0000e-05 - 196ms/epoch - 4ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0553 - val_loss: 0.5359 - val_mse: 0.5359 - val_mae: 0.6848 - lr: 1.0000e-05 - 194ms/epoch - 4ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0589 - val_loss: 0.5355 - val_mse: 0.5355 - val_mae: 0.6845 - lr: 1.0000e-05 - 232ms/epoch - 5ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0581 - val_loss: 0.5351 - val_mse: 0.5351 - val_mae: 0.6843 - lr: 1.0000e-05 - 273ms/epoch - 6ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0052 - mse: 0.0052 - mae: 0.0561 - val_loss: 0.5346 - val_mse: 0.5346 - val_mae: 0.6839 - lr: 1.0000e-05 - 220ms/epoch - 5ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0571 - val_loss: 0.5329 - val_mse: 0.5329 - val_mae: 0.6828 - lr: 1.0000e-05 - 221ms/epoch - 5ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0589 - val_loss: 0.5332 - val_mse: 0.5332 - val_mae: 0.6830 - lr: 1.0000e-05 - 232ms/epoch - 5ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0054 - mse: 0.0054 - mae: 0.0567 - val_loss: 0.5325 - val_mse: 0.5325 - val_mae: 0.6826 - lr: 1.0000e-05 - 226ms/epoch - 5ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0585 - val_loss: 0.5329 - val_mse: 0.5329 - val_mae: 0.6829 - lr: 1.0000e-05 - 267ms/epoch - 6ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0058 - mse: 0.0058 - mae: 0.0585 - val_loss: 0.5338 - val_mse: 0.5338 - val_mae: 0.6836 - lr: 1.0000e-05 - 233ms/epoch - 5ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0549 - val_loss: 0.5329 - val_mse: 0.5329 - val_mae: 0.6829 - lr: 1.0000e-05 - 216ms/epoch - 5ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0049 - mse: 0.0049 - mae: 0.0540 - val_loss: 0.5328 - val_mse: 0.5328 - val_mae: 0.6828 - lr: 1.0000e-05 - 236ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0570 - val_loss: 0.5322 - val_mse: 0.5322 - val_mae: 0.6825 - lr: 1.0000e-05 - 217ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0589 - val_loss: 0.5294 - val_mse: 0.5294 - val_mae: 0.6806 - lr: 1.0000e-05 - 243ms/epoch - 5ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0584 - val_loss: 0.5271 - val_mse: 0.5271 - val_mae: 0.6790 - lr: 1.0000e-05 - 204ms/epoch - 4ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0583 - val_loss: 0.5264 - val_mse: 0.5264 - val_mae: 0.6785 - lr: 1.0000e-05 - 228ms/epoch - 5ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0574 - val_loss: 0.5251 - val_mse: 0.5251 - val_mae: 0.6776 - lr: 1.0000e-05 - 204ms/epoch - 4ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0565 - val_loss: 0.5230 - val_mse: 0.5230 - val_mae: 0.6762 - lr: 1.0000e-05 - 216ms/epoch - 5ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0583 - val_loss: 0.5216 - val_mse: 0.5216 - val_mae: 0.6753 - lr: 1.0000e-05 - 237ms/epoch - 5ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0056 - mse: 0.0056 - mae: 0.0564 - val_loss: 0.5218 - val_mse: 0.5218 - val_mae: 0.6754 - lr: 1.0000e-05 - 206ms/epoch - 4ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0580 - val_loss: 0.5229 - val_mse: 0.5229 - val_mae: 0.6762 - lr: 1.0000e-05 - 215ms/epoch - 4ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0586 - val_loss: 0.5221 - val_mse: 0.5221 - val_mae: 0.6757 - lr: 1.0000e-05 - 261ms/epoch - 5ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0052 - mse: 0.0052 - mae: 0.0557 - val_loss: 0.5206 - val_mse: 0.5206 - val_mae: 0.6747 - lr: 1.0000e-05 - 237ms/epoch - 5ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0572 - val_loss: 0.5215 - val_mse: 0.5215 - val_mae: 0.6754 - lr: 1.0000e-05 - 228ms/epoch - 5ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0594 - val_loss: 0.5210 - val_mse: 0.5210 - val_mae: 0.6750 - lr: 1.0000e-05 - 249ms/epoch - 5ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0058 - mse: 0.0058 - mae: 0.0574 - val_loss: 0.5210 - val_mse: 0.5210 - val_mae: 0.6751 - lr: 1.0000e-05 - 234ms/epoch - 5ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0052 - mse: 0.0052 - mae: 0.0553 - val_loss: 0.5204 - val_mse: 0.5204 - val_mae: 0.6747 - lr: 1.0000e-05 - 192ms/epoch - 4ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0561 - val_loss: 0.5199 - val_mse: 0.5199 - val_mae: 0.6744 - lr: 1.0000e-05 - 196ms/epoch - 4ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0585 - val_loss: 0.5201 - val_mse: 0.5201 - val_mae: 0.6746 - lr: 1.0000e-05 - 206ms/epoch - 4ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.47229
48/48 - 0s - loss: 0.0052 - mse: 0.0052 - mae: 0.0560 - val_loss: 0.5195 - val_mse: 0.5195 - val_mae: 0.6742 - lr: 1.0000e-05 - 225ms/epoch - 5ms/step
Epoch 00053: early stopping
SMA
Prediction vs Close: 50.0% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 133.6684179910327
RMSE: 11.561505870388714
MAPE: 10.289389775089397
EMA
EMA([input_arrays], [timeperiod=30])
Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
51
Working on EMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.778, Time=3.24 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.589, Time=4.62 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-14606.447, Time=5.96 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.589, Time=7.07 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-15343.613, Time=9.68 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-15047.583, Time=12.72 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-16858.964, Time=11.56 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=-17024.022, Time=6.23 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=-16998.618, Time=3.56 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=-17081.451, Time=6.80 sec
ARIMA(3,3,2)(0,0,0)[0] : AIC=inf, Time=17.00 sec
/usr/local/lib/python3.7/dist-packages/statsmodels/tsa/statespace/sarimax.py:1906: RuntimeWarning: divide by zero encountered in reciprocal return np.roots(self.polynomial_reduced_ma)**-1
ARIMA(2,3,2)(0,0,0)[0] : AIC=-16997.990, Time=3.75 sec
ARIMA(3,3,1)(0,0,0)[0] intercept : AIC=-16992.667, Time=4.25 sec
Best model: ARIMA(3,3,1)(0,0,0)[0]
Total fit time: 96.460 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 1) Log Likelihood 8569.726
Date: Sun, 12 Dec 2021 AIC -17081.451
Time: 17:11:15 BIC -16945.417
Sample: 0 HQIC -17029.208
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -2.316e-10 9.89e-05 -2.34e-06 1.000 -0.000 0.000
x2 -2.309e-10 9.88e-05 -2.34e-06 1.000 -0.000 0.000
x3 -2.325e-10 9.91e-05 -2.35e-06 1.000 -0.000 0.000
x4 1.0000 9.9e-05 1.01e+04 0.000 1.000 1.000
x5 -2.108e-10 9.43e-05 -2.24e-06 1.000 -0.000 0.000
x6 -7.997e-10 0.000 -4.63e-06 1.000 -0.000 0.000
x7 -2.295e-10 9.85e-05 -2.33e-06 1.000 -0.000 0.000
x8 -2.244e-10 9.74e-05 -2.3e-06 1.000 -0.000 0.000
x9 -1.166e-11 1.98e-05 -5.9e-07 1.000 -3.87e-05 3.87e-05
x10 -4.454e-11 4.19e-05 -1.06e-06 1.000 -8.22e-05 8.22e-05
x11 -2.219e-10 9.68e-05 -2.29e-06 1.000 -0.000 0.000
x12 -2.264e-10 9.8e-05 -2.31e-06 1.000 -0.000 0.000
x13 -2.315e-10 9.89e-05 -2.34e-06 1.000 -0.000 0.000
x14 -1.767e-09 0.000 -6.47e-06 1.000 -0.001 0.001
x15 -2.096e-10 9.38e-05 -2.23e-06 1.000 -0.000 0.000
x16 -5.257e-10 0.000 -3.5e-06 1.000 -0.000 0.000
x17 -2.143e-10 9.53e-05 -2.25e-06 1.000 -0.000 0.000
x18 -3.776e-11 3.61e-05 -1.05e-06 1.000 -7.08e-05 7.08e-05
x19 -2.52e-10 0.000 -2.41e-06 1.000 -0.000 0.000
x20 -2.417e-10 9.51e-05 -2.54e-06 1.000 -0.000 0.000
x21 -3.16e-09 0.000 -8.64e-06 1.000 -0.001 0.001
x22 -2.955e-09 0.000 -8.32e-06 1.000 -0.001 0.001
x23 -1.664e-09 0.000 -6.29e-06 1.000 -0.001 0.001
x24 -1.568e-09 0.000 -6.07e-06 1.000 -0.001 0.001
ar.L1 -0.4923 1.2e-09 -4.09e+08 0.000 -0.492 -0.492
ar.L2 -0.1923 7e-10 -2.75e+08 0.000 -0.192 -0.192
ar.L3 -0.0461 3.32e-10 -1.39e+08 0.000 -0.046 -0.046
ma.L1 -0.7077 2.73e-09 -2.59e+08 0.000 -0.708 -0.708
sigma2 8.99e-11 6.96e-11 1.291 0.197 -4.66e-11 2.26e-10
===================================================================================
Ljung-Box (L1) (Q): 55.51 Jarque-Bera (JB): 4268313.90
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 5.44
Prob(H) (two-sided): 0.00 Kurtosis: 359.56
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.36e+28. Standard errors may be unstable.
ARIMA order: (3, 3, 1)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.14368, saving model to LSTM7.h5
16/16 - 2s - loss: 0.1485 - mse: 0.1485 - mae: 0.2985 - val_loss: 0.1437 - val_mse: 0.1437 - val_mae: 0.3319 - lr: 0.0010 - 2s/epoch - 145ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0519 - mse: 0.0519 - mae: 0.1969 - val_loss: 0.1677 - val_mse: 0.1677 - val_mae: 0.3644 - lr: 0.0010 - 77ms/epoch - 5ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0194 - mse: 0.0194 - mae: 0.1138 - val_loss: 0.1986 - val_mse: 0.1986 - val_mae: 0.4015 - lr: 0.0010 - 78ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0142 - mse: 0.0142 - mae: 0.0937 - val_loss: 0.1694 - val_mse: 0.1694 - val_mae: 0.3635 - lr: 0.0010 - 94ms/epoch - 6ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0140 - mse: 0.0140 - mae: 0.0926 - val_loss: 0.1751 - val_mse: 0.1751 - val_mae: 0.3703 - lr: 0.0010 - 77ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00006: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0117 - mse: 0.0117 - mae: 0.0867 - val_loss: 0.1673 - val_mse: 0.1673 - val_mae: 0.3603 - lr: 0.0010 - 91ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0105 - mse: 0.0105 - mae: 0.0796 - val_loss: 0.1668 - val_mse: 0.1668 - val_mae: 0.3596 - lr: 1.0000e-04 - 83ms/epoch - 5ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0108 - mse: 0.0108 - mae: 0.0825 - val_loss: 0.1656 - val_mse: 0.1656 - val_mae: 0.3579 - lr: 1.0000e-04 - 93ms/epoch - 6ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0109 - mse: 0.0109 - mae: 0.0837 - val_loss: 0.1643 - val_mse: 0.1643 - val_mae: 0.3561 - lr: 1.0000e-04 - 94ms/epoch - 6ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0111 - mse: 0.0111 - mae: 0.0829 - val_loss: 0.1626 - val_mse: 0.1626 - val_mae: 0.3538 - lr: 1.0000e-04 - 97ms/epoch - 6ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00011: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0108 - mse: 0.0108 - mae: 0.0824 - val_loss: 0.1614 - val_mse: 0.1614 - val_mae: 0.3522 - lr: 1.0000e-04 - 96ms/epoch - 6ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0101 - mse: 0.0101 - mae: 0.0805 - val_loss: 0.1614 - val_mse: 0.1614 - val_mae: 0.3522 - lr: 1.0000e-05 - 79ms/epoch - 5ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0104 - mse: 0.0104 - mae: 0.0802 - val_loss: 0.1614 - val_mse: 0.1614 - val_mae: 0.3522 - lr: 1.0000e-05 - 84ms/epoch - 5ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0103 - mse: 0.0103 - mae: 0.0815 - val_loss: 0.1613 - val_mse: 0.1613 - val_mae: 0.3520 - lr: 1.0000e-05 - 89ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0107 - mse: 0.0107 - mae: 0.0822 - val_loss: 0.1613 - val_mse: 0.1613 - val_mae: 0.3521 - lr: 1.0000e-05 - 88ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00016: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0103 - mse: 0.0103 - mae: 0.0817 - val_loss: 0.1613 - val_mse: 0.1613 - val_mae: 0.3520 - lr: 1.0000e-05 - 81ms/epoch - 5ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0105 - mse: 0.0105 - mae: 0.0813 - val_loss: 0.1611 - val_mse: 0.1611 - val_mae: 0.3517 - lr: 1.0000e-05 - 80ms/epoch - 5ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0098 - mse: 0.0098 - mae: 0.0789 - val_loss: 0.1612 - val_mse: 0.1612 - val_mae: 0.3518 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0102 - mse: 0.0102 - mae: 0.0783 - val_loss: 0.1611 - val_mse: 0.1611 - val_mae: 0.3517 - lr: 1.0000e-05 - 81ms/epoch - 5ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0105 - mse: 0.0105 - mae: 0.0817 - val_loss: 0.1610 - val_mse: 0.1610 - val_mae: 0.3516 - lr: 1.0000e-05 - 83ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0094 - mse: 0.0094 - mae: 0.0782 - val_loss: 0.1611 - val_mse: 0.1611 - val_mae: 0.3517 - lr: 1.0000e-05 - 96ms/epoch - 6ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0115 - mse: 0.0115 - mae: 0.0851 - val_loss: 0.1610 - val_mse: 0.1610 - val_mae: 0.3515 - lr: 1.0000e-05 - 96ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0105 - mse: 0.0105 - mae: 0.0817 - val_loss: 0.1608 - val_mse: 0.1608 - val_mae: 0.3513 - lr: 1.0000e-05 - 92ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0105 - mse: 0.0105 - mae: 0.0819 - val_loss: 0.1608 - val_mse: 0.1608 - val_mae: 0.3513 - lr: 1.0000e-05 - 81ms/epoch - 5ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0097 - mse: 0.0097 - mae: 0.0776 - val_loss: 0.1607 - val_mse: 0.1607 - val_mae: 0.3512 - lr: 1.0000e-05 - 82ms/epoch - 5ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0098 - mse: 0.0098 - mae: 0.0787 - val_loss: 0.1608 - val_mse: 0.1608 - val_mae: 0.3513 - lr: 1.0000e-05 - 82ms/epoch - 5ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0112 - mse: 0.0112 - mae: 0.0834 - val_loss: 0.1607 - val_mse: 0.1607 - val_mae: 0.3511 - lr: 1.0000e-05 - 81ms/epoch - 5ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0109 - mse: 0.0109 - mae: 0.0825 - val_loss: 0.1605 - val_mse: 0.1605 - val_mae: 0.3509 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0106 - mse: 0.0106 - mae: 0.0821 - val_loss: 0.1603 - val_mse: 0.1603 - val_mae: 0.3507 - lr: 1.0000e-05 - 84ms/epoch - 5ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0100 - mse: 0.0100 - mae: 0.0795 - val_loss: 0.1603 - val_mse: 0.1603 - val_mae: 0.3506 - lr: 1.0000e-05 - 83ms/epoch - 5ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0106 - mse: 0.0106 - mae: 0.0820 - val_loss: 0.1600 - val_mse: 0.1600 - val_mae: 0.3502 - lr: 1.0000e-05 - 93ms/epoch - 6ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0101 - mse: 0.0101 - mae: 0.0802 - val_loss: 0.1598 - val_mse: 0.1598 - val_mae: 0.3499 - lr: 1.0000e-05 - 97ms/epoch - 6ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0099 - mse: 0.0099 - mae: 0.0782 - val_loss: 0.1596 - val_mse: 0.1596 - val_mae: 0.3496 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0113 - mse: 0.0113 - mae: 0.0835 - val_loss: 0.1594 - val_mse: 0.1594 - val_mae: 0.3493 - lr: 1.0000e-05 - 83ms/epoch - 5ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0104 - mse: 0.0104 - mae: 0.0824 - val_loss: 0.1592 - val_mse: 0.1592 - val_mae: 0.3492 - lr: 1.0000e-05 - 84ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0099 - mse: 0.0099 - mae: 0.0781 - val_loss: 0.1592 - val_mse: 0.1592 - val_mae: 0.3491 - lr: 1.0000e-05 - 86ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0092 - mse: 0.0092 - mae: 0.0759 - val_loss: 0.1590 - val_mse: 0.1590 - val_mae: 0.3489 - lr: 1.0000e-05 - 78ms/epoch - 5ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0114 - mse: 0.0114 - mae: 0.0852 - val_loss: 0.1589 - val_mse: 0.1589 - val_mae: 0.3487 - lr: 1.0000e-05 - 93ms/epoch - 6ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0108 - mse: 0.0108 - mae: 0.0823 - val_loss: 0.1587 - val_mse: 0.1587 - val_mae: 0.3484 - lr: 1.0000e-05 - 86ms/epoch - 5ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0100 - mse: 0.0100 - mae: 0.0784 - val_loss: 0.1589 - val_mse: 0.1589 - val_mae: 0.3486 - lr: 1.0000e-05 - 92ms/epoch - 6ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0111 - mse: 0.0111 - mae: 0.0829 - val_loss: 0.1588 - val_mse: 0.1588 - val_mae: 0.3486 - lr: 1.0000e-05 - 83ms/epoch - 5ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0103 - mse: 0.0103 - mae: 0.0803 - val_loss: 0.1589 - val_mse: 0.1589 - val_mae: 0.3487 - lr: 1.0000e-05 - 82ms/epoch - 5ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0106 - mse: 0.0106 - mae: 0.0809 - val_loss: 0.1587 - val_mse: 0.1587 - val_mae: 0.3484 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0107 - mse: 0.0107 - mae: 0.0813 - val_loss: 0.1585 - val_mse: 0.1585 - val_mae: 0.3482 - lr: 1.0000e-05 - 89ms/epoch - 6ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0088 - mse: 0.0088 - mae: 0.0756 - val_loss: 0.1585 - val_mse: 0.1585 - val_mae: 0.3481 - lr: 1.0000e-05 - 82ms/epoch - 5ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0107 - mse: 0.0107 - mae: 0.0827 - val_loss: 0.1584 - val_mse: 0.1584 - val_mae: 0.3479 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0102 - mse: 0.0102 - mae: 0.0811 - val_loss: 0.1582 - val_mse: 0.1582 - val_mae: 0.3477 - lr: 1.0000e-05 - 100ms/epoch - 6ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0097 - mse: 0.0097 - mae: 0.0792 - val_loss: 0.1579 - val_mse: 0.1579 - val_mae: 0.3473 - lr: 1.0000e-05 - 84ms/epoch - 5ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0102 - mse: 0.0102 - mae: 0.0811 - val_loss: 0.1575 - val_mse: 0.1575 - val_mae: 0.3468 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0107 - mse: 0.0107 - mae: 0.0818 - val_loss: 0.1573 - val_mse: 0.1573 - val_mae: 0.3465 - lr: 1.0000e-05 - 92ms/epoch - 6ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.14368
16/16 - 0s - loss: 0.0102 - mse: 0.0102 - mae: 0.0798 - val_loss: 0.1572 - val_mse: 0.1572 - val_mae: 0.3463 - lr: 1.0000e-05 - 86ms/epoch - 5ms/step
Epoch 00051: early stopping
SMA
Prediction vs Close: 50.0% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 133.6684179910327
RMSE: 11.561505870388714
MAPE: 10.289389775089397
EMA
Prediction vs Close: 50.0% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 34.52931938659473
RMSE: 5.876165364129462
MAPE: 4.852473639818076
WMA
WMA([input_arrays], [timeperiod=30])
Weighted Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
49
Working on WMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.780, Time=2.89 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.589, Time=4.60 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-16789.784, Time=12.16 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.589, Time=7.24 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16919.987, Time=9.55 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-14616.097, Time=12.57 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17225.955, Time=18.58 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-14562.589, Time=9.75 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-15582.364, Time=19.25 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-12043.670, Time=36.40 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 133.001 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8639.977
Date: Sun, 12 Dec 2021 AIC -17225.955
Time: 17:22:29 BIC -17099.302
Sample: 0 HQIC -17177.315
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -4.802e-09 4.51e-06 -0.001 0.999 -8.84e-06 8.83e-06
x2 -4.783e-09 4.5e-06 -0.001 0.999 -8.83e-06 8.82e-06
x3 -4.811e-09 4.51e-06 -0.001 0.999 -8.85e-06 8.84e-06
x4 1.0000 4.51e-06 2.22e+05 0.000 1.000 1.000
x5 -4.353e-09 4.3e-06 -0.001 0.999 -8.43e-06 8.42e-06
x6 -1.569e-08 7.54e-06 -0.002 0.998 -1.48e-05 1.48e-05
x7 -4.75e-09 4.49e-06 -0.001 0.999 -8.8e-06 8.79e-06
x8 -4.628e-09 4.43e-06 -0.001 0.999 -8.69e-06 8.69e-06
x9 -4.733e-10 1.16e-06 -0.000 1.000 -2.27e-06 2.27e-06
x10 -7.88e-10 1.8e-06 -0.000 1.000 -3.52e-06 3.52e-06
x11 -4.609e-09 4.42e-06 -0.001 0.999 -8.68e-06 8.67e-06
x12 -4.607e-09 4.42e-06 -0.001 0.999 -8.68e-06 8.67e-06
x13 -4.792e-09 4.51e-06 -0.001 0.999 -8.84e-06 8.83e-06
x14 -3.777e-08 1.24e-05 -0.003 0.998 -2.44e-05 2.44e-05
x15 -3.99e-09 4.12e-06 -0.001 0.999 -8.08e-06 8.07e-06
x16 -1.309e-08 7.41e-06 -0.002 0.999 -1.45e-05 1.45e-05
x17 -4.789e-09 4.51e-06 -0.001 0.999 -8.85e-06 8.84e-06
x18 -2.665e-10 9.77e-07 -0.000 1.000 -1.92e-06 1.92e-06
x19 -4.919e-09 4.56e-06 -0.001 0.999 -8.94e-06 8.93e-06
x20 -4e-10 9.58e-07 -0.000 1.000 -1.88e-06 1.88e-06
x21 -6.782e-08 1.67e-05 -0.004 0.997 -3.27e-05 3.26e-05
x22 -6.03e-08 1.58e-05 -0.004 0.997 -3.09e-05 3.08e-05
x23 -3.157e-08 1.14e-05 -0.003 0.998 -2.23e-05 2.23e-05
x24 -3.671e-08 1.23e-05 -0.003 0.998 -2.41e-05 2.41e-05
ma.L1 -1.3901 5.58e-10 -2.49e+09 0.000 -1.390 -1.390
ma.L2 0.4033 5.75e-10 7.02e+08 0.000 0.403 0.403
sigma2 7.525e-11 6.92e-11 1.088 0.277 -6.03e-11 2.11e-10
===================================================================================
Ljung-Box (L1) (Q): 69.18 Jarque-Bera (JB): 6366427.21
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 12.29
Prob(H) (two-sided): 0.00 Kurtosis: 437.97
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 2.29e+25. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.07482, saving model to LSTM7.h5
17/17 - 3s - loss: 0.0780 - mse: 0.0780 - mae: 0.2303 - val_loss: 0.0748 - val_mse: 0.0748 - val_mae: 0.2408 - lr: 0.0010 - 3s/epoch - 159ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.07482 to 0.06158, saving model to LSTM7.h5
17/17 - 0s - loss: 0.0356 - mse: 0.0356 - mae: 0.1480 - val_loss: 0.0616 - val_mse: 0.0616 - val_mae: 0.2091 - lr: 0.0010 - 91ms/epoch - 5ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0187 - mse: 0.0187 - mae: 0.1092 - val_loss: 0.0764 - val_mse: 0.0764 - val_mae: 0.2089 - lr: 0.0010 - 87ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0161 - mse: 0.0161 - mae: 0.1015 - val_loss: 0.0864 - val_mse: 0.0864 - val_mae: 0.2191 - lr: 0.0010 - 99ms/epoch - 6ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0148 - mse: 0.0148 - mae: 0.0954 - val_loss: 0.0902 - val_mse: 0.0902 - val_mae: 0.2231 - lr: 0.0010 - 86ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0115 - mse: 0.0115 - mae: 0.0835 - val_loss: 0.1079 - val_mse: 0.1079 - val_mae: 0.2441 - lr: 0.0010 - 87ms/epoch - 5ms/step
Epoch 7/500
Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00007: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0102 - mse: 0.0102 - mae: 0.0798 - val_loss: 0.1203 - val_mse: 0.1203 - val_mae: 0.2608 - lr: 0.0010 - 102ms/epoch - 6ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0098 - mse: 0.0098 - mae: 0.0778 - val_loss: 0.1177 - val_mse: 0.1177 - val_mae: 0.2571 - lr: 1.0000e-04 - 90ms/epoch - 5ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0104 - mse: 0.0104 - mae: 0.0799 - val_loss: 0.1162 - val_mse: 0.1162 - val_mae: 0.2550 - lr: 1.0000e-04 - 84ms/epoch - 5ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0093 - mse: 0.0093 - mae: 0.0763 - val_loss: 0.1157 - val_mse: 0.1157 - val_mae: 0.2544 - lr: 1.0000e-04 - 83ms/epoch - 5ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0094 - mse: 0.0094 - mae: 0.0765 - val_loss: 0.1159 - val_mse: 0.1159 - val_mae: 0.2546 - lr: 1.0000e-04 - 95ms/epoch - 6ms/step
Epoch 12/500
Epoch 00012: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00012: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0091 - mse: 0.0091 - mae: 0.0749 - val_loss: 0.1178 - val_mse: 0.1178 - val_mae: 0.2573 - lr: 1.0000e-04 - 91ms/epoch - 5ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0095 - mse: 0.0095 - mae: 0.0771 - val_loss: 0.1178 - val_mse: 0.1178 - val_mae: 0.2573 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0090 - mse: 0.0090 - mae: 0.0754 - val_loss: 0.1180 - val_mse: 0.1180 - val_mae: 0.2575 - lr: 1.0000e-05 - 89ms/epoch - 5ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0100 - mse: 0.0100 - mae: 0.0782 - val_loss: 0.1178 - val_mse: 0.1178 - val_mae: 0.2573 - lr: 1.0000e-05 - 96ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0092 - mse: 0.0092 - mae: 0.0743 - val_loss: 0.1177 - val_mse: 0.1177 - val_mae: 0.2571 - lr: 1.0000e-05 - 90ms/epoch - 5ms/step
Epoch 17/500
Epoch 00017: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00017: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0089 - mse: 0.0089 - mae: 0.0744 - val_loss: 0.1176 - val_mse: 0.1176 - val_mae: 0.2570 - lr: 1.0000e-05 - 98ms/epoch - 6ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0083 - mse: 0.0083 - mae: 0.0713 - val_loss: 0.1176 - val_mse: 0.1176 - val_mae: 0.2570 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0084 - mse: 0.0084 - mae: 0.0713 - val_loss: 0.1175 - val_mse: 0.1175 - val_mae: 0.2569 - lr: 1.0000e-05 - 83ms/epoch - 5ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0090 - mse: 0.0090 - mae: 0.0743 - val_loss: 0.1175 - val_mse: 0.1175 - val_mae: 0.2569 - lr: 1.0000e-05 - 106ms/epoch - 6ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0091 - mse: 0.0091 - mae: 0.0741 - val_loss: 0.1176 - val_mse: 0.1176 - val_mae: 0.2570 - lr: 1.0000e-05 - 86ms/epoch - 5ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0092 - mse: 0.0092 - mae: 0.0756 - val_loss: 0.1177 - val_mse: 0.1177 - val_mae: 0.2571 - lr: 1.0000e-05 - 82ms/epoch - 5ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0088 - mse: 0.0088 - mae: 0.0749 - val_loss: 0.1179 - val_mse: 0.1179 - val_mae: 0.2575 - lr: 1.0000e-05 - 84ms/epoch - 5ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0097 - mse: 0.0097 - mae: 0.0780 - val_loss: 0.1182 - val_mse: 0.1182 - val_mae: 0.2578 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0085 - mse: 0.0085 - mae: 0.0726 - val_loss: 0.1183 - val_mse: 0.1183 - val_mae: 0.2580 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0088 - mse: 0.0088 - mae: 0.0724 - val_loss: 0.1183 - val_mse: 0.1183 - val_mae: 0.2580 - lr: 1.0000e-05 - 86ms/epoch - 5ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0090 - mse: 0.0090 - mae: 0.0747 - val_loss: 0.1184 - val_mse: 0.1184 - val_mae: 0.2581 - lr: 1.0000e-05 - 92ms/epoch - 5ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0089 - mse: 0.0089 - mae: 0.0738 - val_loss: 0.1183 - val_mse: 0.1183 - val_mae: 0.2580 - lr: 1.0000e-05 - 92ms/epoch - 5ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0087 - mse: 0.0087 - mae: 0.0737 - val_loss: 0.1185 - val_mse: 0.1185 - val_mae: 0.2583 - lr: 1.0000e-05 - 96ms/epoch - 6ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0094 - mse: 0.0094 - mae: 0.0751 - val_loss: 0.1183 - val_mse: 0.1183 - val_mae: 0.2580 - lr: 1.0000e-05 - 83ms/epoch - 5ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0091 - mse: 0.0091 - mae: 0.0750 - val_loss: 0.1183 - val_mse: 0.1183 - val_mae: 0.2581 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0093 - mse: 0.0093 - mae: 0.0762 - val_loss: 0.1184 - val_mse: 0.1184 - val_mae: 0.2581 - lr: 1.0000e-05 - 89ms/epoch - 5ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0087 - mse: 0.0087 - mae: 0.0751 - val_loss: 0.1184 - val_mse: 0.1184 - val_mae: 0.2582 - lr: 1.0000e-05 - 80ms/epoch - 5ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0083 - mse: 0.0083 - mae: 0.0715 - val_loss: 0.1186 - val_mse: 0.1186 - val_mae: 0.2585 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0082 - mse: 0.0082 - mae: 0.0715 - val_loss: 0.1187 - val_mse: 0.1187 - val_mae: 0.2586 - lr: 1.0000e-05 - 90ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0091 - mse: 0.0091 - mae: 0.0748 - val_loss: 0.1187 - val_mse: 0.1187 - val_mae: 0.2587 - lr: 1.0000e-05 - 107ms/epoch - 6ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0084 - mse: 0.0084 - mae: 0.0719 - val_loss: 0.1188 - val_mse: 0.1188 - val_mae: 0.2588 - lr: 1.0000e-05 - 92ms/epoch - 5ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0092 - mse: 0.0092 - mae: 0.0743 - val_loss: 0.1188 - val_mse: 0.1188 - val_mae: 0.2589 - lr: 1.0000e-05 - 80ms/epoch - 5ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0080 - mse: 0.0080 - mae: 0.0710 - val_loss: 0.1189 - val_mse: 0.1189 - val_mae: 0.2589 - lr: 1.0000e-05 - 96ms/epoch - 6ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0089 - mse: 0.0089 - mae: 0.0744 - val_loss: 0.1188 - val_mse: 0.1188 - val_mae: 0.2588 - lr: 1.0000e-05 - 93ms/epoch - 5ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0088 - mse: 0.0088 - mae: 0.0730 - val_loss: 0.1186 - val_mse: 0.1186 - val_mae: 0.2585 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0087 - mse: 0.0087 - mae: 0.0733 - val_loss: 0.1186 - val_mse: 0.1186 - val_mae: 0.2585 - lr: 1.0000e-05 - 87ms/epoch - 5ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0087 - mse: 0.0087 - mae: 0.0738 - val_loss: 0.1186 - val_mse: 0.1186 - val_mae: 0.2586 - lr: 1.0000e-05 - 106ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0083 - mse: 0.0083 - mae: 0.0711 - val_loss: 0.1190 - val_mse: 0.1190 - val_mae: 0.2591 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0088 - mse: 0.0088 - mae: 0.0742 - val_loss: 0.1188 - val_mse: 0.1188 - val_mae: 0.2589 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0093 - mse: 0.0093 - mae: 0.0755 - val_loss: 0.1187 - val_mse: 0.1187 - val_mae: 0.2587 - lr: 1.0000e-05 - 94ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0081 - mse: 0.0081 - mae: 0.0695 - val_loss: 0.1188 - val_mse: 0.1188 - val_mae: 0.2588 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0088 - mse: 0.0088 - mae: 0.0744 - val_loss: 0.1189 - val_mse: 0.1189 - val_mae: 0.2590 - lr: 1.0000e-05 - 89ms/epoch - 5ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0078 - mse: 0.0078 - mae: 0.0708 - val_loss: 0.1194 - val_mse: 0.1194 - val_mae: 0.2597 - lr: 1.0000e-05 - 79ms/epoch - 5ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0083 - mse: 0.0083 - mae: 0.0723 - val_loss: 0.1195 - val_mse: 0.1195 - val_mae: 0.2598 - lr: 1.0000e-05 - 91ms/epoch - 5ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0085 - mse: 0.0085 - mae: 0.0725 - val_loss: 0.1197 - val_mse: 0.1197 - val_mae: 0.2601 - lr: 1.0000e-05 - 83ms/epoch - 5ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.06158
17/17 - 0s - loss: 0.0085 - mse: 0.0085 - mae: 0.0720 - val_loss: 0.1199 - val_mse: 0.1199 - val_mae: 0.2605 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 00052: early stopping
SMA
Prediction vs Close: 50.0% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 133.6684179910327
RMSE: 11.561505870388714
MAPE: 10.289389775089397
EMA
Prediction vs Close: 50.0% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 34.52931938659473
RMSE: 5.876165364129462
MAPE: 4.852473639818076
WMA
Prediction vs Close: 51.49% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 36.74468487644727
RMSE: 6.061739426637149
MAPE: 4.85767480758183
DEMA
DEMA([input_arrays], [timeperiod=30])
Double Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
89
Working on DEMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.785, Time=3.06 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.588, Time=4.53 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-15575.689, Time=8.72 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.588, Time=6.95 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16714.796, Time=8.85 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-15610.140, Time=10.53 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17225.835, Time=22.19 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-14562.588, Time=9.30 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-16751.951, Time=20.80 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-11788.089, Time=30.48 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 125.433 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8639.917
Date: Sun, 12 Dec 2021 AIC -17225.835
Time: 17:29:03 BIC -17099.182
Sample: 0 HQIC -17177.195
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -5.894e-09 3.61e-06 -0.002 0.999 -7.09e-06 7.08e-06
x2 -5.93e-09 3.63e-06 -0.002 0.999 -7.11e-06 7.1e-06
x3 -5.905e-09 3.62e-06 -0.002 0.999 -7.1e-06 7.09e-06
x4 1.0000 3.62e-06 2.76e+05 0.000 1.000 1.000
x5 -5.457e-09 3.48e-06 -0.002 0.999 -6.83e-06 6.82e-06
x6 -3.019e-08 7.72e-06 -0.004 0.997 -1.52e-05 1.51e-05
x7 -5.87e-09 3.61e-06 -0.002 0.999 -7.08e-06 7.07e-06
x8 -5.809e-09 3.59e-06 -0.002 0.999 -7.05e-06 7.04e-06
x9 -9.293e-11 9.83e-08 -0.001 0.999 -1.93e-07 1.93e-07
x10 -2.793e-09 2.47e-06 -0.001 0.999 -4.84e-06 4.84e-06
x11 -6.095e-09 3.68e-06 -0.002 0.999 -7.21e-06 7.2e-06
x12 -5.478e-09 3.49e-06 -0.002 0.999 -6.85e-06 6.84e-06
x13 -5.91e-09 3.62e-06 -0.002 0.999 -7.1e-06 7.09e-06
x14 -4.085e-08 9.35e-06 -0.004 0.997 -1.84e-05 1.83e-05
x15 -5.93e-09 3.63e-06 -0.002 0.999 -7.12e-06 7.11e-06
x16 -1.618e-09 1.92e-06 -0.001 0.999 -3.76e-06 3.75e-06
x17 -5.076e-09 3.37e-06 -0.002 0.999 -6.6e-06 6.59e-06
x18 -1.377e-08 5.5e-06 -0.003 0.998 -1.08e-05 1.08e-05
x19 -6.135e-09 3.69e-06 -0.002 0.999 -7.23e-06 7.22e-06
x20 -1.018e-08 4.43e-06 -0.002 0.998 -8.68e-06 8.66e-06
x21 -6.911e-08 1.21e-05 -0.006 0.995 -2.39e-05 2.37e-05
x22 -5.656e-08 1.1e-05 -0.005 0.996 -2.16e-05 2.15e-05
x23 -5.355e-08 1.07e-05 -0.005 0.996 -2.1e-05 2.09e-05
x24 -3.636e-08 8.85e-06 -0.004 0.997 -1.74e-05 1.73e-05
ma.L1 -1.3899 4.86e-11 -2.86e+10 0.000 -1.390 -1.390
ma.L2 0.4032 4.6e-11 8.76e+09 0.000 0.403 0.403
sigma2 7.526e-11 6.92e-11 1.088 0.277 -6.03e-11 2.11e-10
===================================================================================
Ljung-Box (L1) (Q): 69.65 Jarque-Bera (JB): 6422892.15
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 12.42
Prob(H) (two-sided): 0.00 Kurtosis: 439.89
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 2.74e+29. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.70173, saving model to LSTM7.h5
10/10 - 2s - loss: 0.8879 - mse: 0.8879 - mae: 0.7859 - val_loss: 0.7017 - val_mse: 0.7017 - val_mae: 0.8156 - lr: 0.0010 - 2s/epoch - 230ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.70173 to 0.38309, saving model to LSTM7.h5
10/10 - 0s - loss: 0.1203 - mse: 0.1203 - mae: 0.2714 - val_loss: 0.3831 - val_mse: 0.3831 - val_mae: 0.5890 - lr: 0.0010 - 67ms/epoch - 7ms/step
Epoch 3/500
Epoch 00003: val_loss improved from 0.38309 to 0.27460, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0877 - mse: 0.0877 - mae: 0.2461 - val_loss: 0.2746 - val_mse: 0.2746 - val_mae: 0.4891 - lr: 0.0010 - 84ms/epoch - 8ms/step
Epoch 4/500
Epoch 00004: val_loss improved from 0.27460 to 0.24714, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0776 - mse: 0.0776 - mae: 0.2356 - val_loss: 0.2471 - val_mse: 0.2471 - val_mae: 0.4620 - lr: 0.0010 - 70ms/epoch - 7ms/step
Epoch 5/500
Epoch 00005: val_loss improved from 0.24714 to 0.23605, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0446 - mse: 0.0446 - mae: 0.1703 - val_loss: 0.2360 - val_mse: 0.2360 - val_mae: 0.4519 - lr: 0.0010 - 82ms/epoch - 8ms/step
Epoch 6/500
Epoch 00006: val_loss improved from 0.23605 to 0.20789, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0371 - mse: 0.0371 - mae: 0.1529 - val_loss: 0.2079 - val_mse: 0.2079 - val_mae: 0.4212 - lr: 0.0010 - 72ms/epoch - 7ms/step
Epoch 7/500
Epoch 00007: val_loss improved from 0.20789 to 0.17846, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0307 - mse: 0.0307 - mae: 0.1320 - val_loss: 0.1785 - val_mse: 0.1785 - val_mae: 0.3859 - lr: 0.0010 - 68ms/epoch - 7ms/step
Epoch 8/500
Epoch 00008: val_loss improved from 0.17846 to 0.15659, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0258 - mse: 0.0258 - mae: 0.1266 - val_loss: 0.1566 - val_mse: 0.1566 - val_mae: 0.3577 - lr: 0.0010 - 80ms/epoch - 8ms/step
Epoch 9/500
Epoch 00009: val_loss improved from 0.15659 to 0.14034, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0228 - mse: 0.0228 - mae: 0.1191 - val_loss: 0.1403 - val_mse: 0.1403 - val_mae: 0.3356 - lr: 0.0010 - 102ms/epoch - 10ms/step
Epoch 10/500
Epoch 00010: val_loss improved from 0.14034 to 0.13114, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0209 - mse: 0.0209 - mae: 0.1142 - val_loss: 0.1311 - val_mse: 0.1311 - val_mae: 0.3230 - lr: 0.0010 - 73ms/epoch - 7ms/step
Epoch 11/500
Epoch 00011: val_loss improved from 0.13114 to 0.11841, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0208 - mse: 0.0208 - mae: 0.1138 - val_loss: 0.1184 - val_mse: 0.1184 - val_mae: 0.3043 - lr: 0.0010 - 73ms/epoch - 7ms/step
Epoch 12/500
Epoch 00012: val_loss improved from 0.11841 to 0.11155, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0183 - mse: 0.0183 - mae: 0.1071 - val_loss: 0.1116 - val_mse: 0.1116 - val_mae: 0.2942 - lr: 0.0010 - 69ms/epoch - 7ms/step
Epoch 13/500
Epoch 00013: val_loss improved from 0.11155 to 0.10249, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0157 - mse: 0.0157 - mae: 0.0967 - val_loss: 0.1025 - val_mse: 0.1025 - val_mae: 0.2799 - lr: 0.0010 - 80ms/epoch - 8ms/step
Epoch 14/500
Epoch 00014: val_loss improved from 0.10249 to 0.09636, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0165 - mse: 0.0165 - mae: 0.1001 - val_loss: 0.0964 - val_mse: 0.0964 - val_mae: 0.2701 - lr: 0.0010 - 88ms/epoch - 9ms/step
Epoch 15/500
Epoch 00015: val_loss improved from 0.09636 to 0.09416, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0141 - mse: 0.0141 - mae: 0.0953 - val_loss: 0.0942 - val_mse: 0.0942 - val_mae: 0.2671 - lr: 0.0010 - 71ms/epoch - 7ms/step
Epoch 16/500
Epoch 00016: val_loss improved from 0.09416 to 0.08719, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0129 - mse: 0.0129 - mae: 0.0894 - val_loss: 0.0872 - val_mse: 0.0872 - val_mae: 0.2548 - lr: 0.0010 - 77ms/epoch - 8ms/step
Epoch 17/500
Epoch 00017: val_loss improved from 0.08719 to 0.08293, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0148 - mse: 0.0148 - mae: 0.0946 - val_loss: 0.0829 - val_mse: 0.0829 - val_mae: 0.2473 - lr: 0.0010 - 78ms/epoch - 8ms/step
Epoch 18/500
Epoch 00018: val_loss improved from 0.08293 to 0.07710, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0130 - mse: 0.0130 - mae: 0.0895 - val_loss: 0.0771 - val_mse: 0.0771 - val_mae: 0.2366 - lr: 0.0010 - 76ms/epoch - 8ms/step
Epoch 19/500
Epoch 00019: val_loss improved from 0.07710 to 0.07451, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0124 - mse: 0.0124 - mae: 0.0873 - val_loss: 0.0745 - val_mse: 0.0745 - val_mae: 0.2321 - lr: 0.0010 - 106ms/epoch - 11ms/step
Epoch 20/500
Epoch 00020: val_loss improved from 0.07451 to 0.07246, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0121 - mse: 0.0121 - mae: 0.0832 - val_loss: 0.0725 - val_mse: 0.0725 - val_mae: 0.2289 - lr: 0.0010 - 91ms/epoch - 9ms/step
Epoch 21/500
Epoch 00021: val_loss improved from 0.07246 to 0.07118, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0115 - mse: 0.0115 - mae: 0.0837 - val_loss: 0.0712 - val_mse: 0.0712 - val_mae: 0.2270 - lr: 0.0010 - 94ms/epoch - 9ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.07118
10/10 - 0s - loss: 0.0112 - mse: 0.0112 - mae: 0.0839 - val_loss: 0.0744 - val_mse: 0.0744 - val_mae: 0.2343 - lr: 0.0010 - 80ms/epoch - 8ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.07118
10/10 - 0s - loss: 0.0113 - mse: 0.0113 - mae: 0.0829 - val_loss: 0.0728 - val_mse: 0.0728 - val_mae: 0.2317 - lr: 0.0010 - 62ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.07118
10/10 - 0s - loss: 0.0105 - mse: 0.0105 - mae: 0.0803 - val_loss: 0.0718 - val_mse: 0.0718 - val_mae: 0.2301 - lr: 0.0010 - 79ms/epoch - 8ms/step
Epoch 25/500
Epoch 00025: val_loss improved from 0.07118 to 0.06944, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0104 - mse: 0.0104 - mae: 0.0796 - val_loss: 0.0694 - val_mse: 0.0694 - val_mae: 0.2258 - lr: 0.0010 - 80ms/epoch - 8ms/step
Epoch 26/500
Epoch 00026: val_loss improved from 0.06944 to 0.06827, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0089 - mse: 0.0089 - mae: 0.0749 - val_loss: 0.0683 - val_mse: 0.0683 - val_mae: 0.2238 - lr: 0.0010 - 72ms/epoch - 7ms/step
Epoch 27/500
Epoch 00027: val_loss improved from 0.06827 to 0.06524, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0086 - mse: 0.0086 - mae: 0.0726 - val_loss: 0.0652 - val_mse: 0.0652 - val_mae: 0.2178 - lr: 0.0010 - 85ms/epoch - 8ms/step
Epoch 28/500
Epoch 00028: val_loss improved from 0.06524 to 0.06351, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0091 - mse: 0.0091 - mae: 0.0749 - val_loss: 0.0635 - val_mse: 0.0635 - val_mae: 0.2144 - lr: 0.0010 - 75ms/epoch - 8ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.06351
10/10 - 0s - loss: 0.0092 - mse: 0.0092 - mae: 0.0758 - val_loss: 0.0643 - val_mse: 0.0643 - val_mae: 0.2168 - lr: 0.0010 - 59ms/epoch - 6ms/step
Epoch 30/500
Epoch 00030: val_loss improved from 0.06351 to 0.06295, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0089 - mse: 0.0089 - mae: 0.0732 - val_loss: 0.0630 - val_mse: 0.0630 - val_mae: 0.2142 - lr: 0.0010 - 87ms/epoch - 9ms/step
Epoch 31/500
Epoch 00031: val_loss improved from 0.06295 to 0.05798, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0081 - mse: 0.0081 - mae: 0.0703 - val_loss: 0.0580 - val_mse: 0.0580 - val_mae: 0.2031 - lr: 0.0010 - 72ms/epoch - 7ms/step
Epoch 32/500
Epoch 00032: val_loss improved from 0.05798 to 0.05263, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0076 - mse: 0.0076 - mae: 0.0692 - val_loss: 0.0526 - val_mse: 0.0526 - val_mae: 0.1907 - lr: 0.0010 - 81ms/epoch - 8ms/step
Epoch 33/500
Epoch 00033: val_loss improved from 0.05263 to 0.05221, saving model to LSTM7.h5
10/10 - 0s - loss: 0.0077 - mse: 0.0077 - mae: 0.0684 - val_loss: 0.0522 - val_mse: 0.0522 - val_mae: 0.1899 - lr: 0.0010 - 79ms/epoch - 8ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0078 - mse: 0.0078 - mae: 0.0678 - val_loss: 0.0534 - val_mse: 0.0534 - val_mae: 0.1927 - lr: 0.0010 - 63ms/epoch - 6ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0072 - mse: 0.0072 - mae: 0.0672 - val_loss: 0.0552 - val_mse: 0.0552 - val_mae: 0.1972 - lr: 0.0010 - 60ms/epoch - 6ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0068 - mse: 0.0068 - mae: 0.0650 - val_loss: 0.0598 - val_mse: 0.0598 - val_mae: 0.2083 - lr: 0.0010 - 55ms/epoch - 6ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0063 - mse: 0.0063 - mae: 0.0629 - val_loss: 0.0640 - val_mse: 0.0640 - val_mae: 0.2176 - lr: 0.0010 - 75ms/epoch - 7ms/step
Epoch 38/500
Epoch 00038: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00038: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0071 - mse: 0.0071 - mae: 0.0655 - val_loss: 0.0587 - val_mse: 0.0587 - val_mae: 0.2060 - lr: 0.0010 - 69ms/epoch - 7ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0065 - mse: 0.0065 - mae: 0.0646 - val_loss: 0.0583 - val_mse: 0.0583 - val_mae: 0.2049 - lr: 1.0000e-04 - 72ms/epoch - 7ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0627 - val_loss: 0.0581 - val_mse: 0.0581 - val_mae: 0.2044 - lr: 1.0000e-04 - 71ms/epoch - 7ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0058 - mse: 0.0058 - mae: 0.0602 - val_loss: 0.0578 - val_mse: 0.0578 - val_mae: 0.2037 - lr: 1.0000e-04 - 71ms/epoch - 7ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0068 - mse: 0.0068 - mae: 0.0651 - val_loss: 0.0577 - val_mse: 0.0577 - val_mae: 0.2037 - lr: 1.0000e-04 - 60ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00043: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0607 - val_loss: 0.0576 - val_mse: 0.0576 - val_mae: 0.2034 - lr: 1.0000e-04 - 59ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0610 - val_loss: 0.0576 - val_mse: 0.0576 - val_mae: 0.2033 - lr: 1.0000e-05 - 71ms/epoch - 7ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0612 - val_loss: 0.0575 - val_mse: 0.0575 - val_mae: 0.2032 - lr: 1.0000e-05 - 75ms/epoch - 7ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0065 - mse: 0.0065 - mae: 0.0629 - val_loss: 0.0575 - val_mse: 0.0575 - val_mae: 0.2031 - lr: 1.0000e-05 - 88ms/epoch - 9ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0627 - val_loss: 0.0574 - val_mse: 0.0574 - val_mae: 0.2029 - lr: 1.0000e-05 - 75ms/epoch - 7ms/step
Epoch 48/500
Epoch 00048: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00048: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0065 - mse: 0.0065 - mae: 0.0622 - val_loss: 0.0574 - val_mse: 0.0574 - val_mae: 0.2029 - lr: 1.0000e-05 - 66ms/epoch - 7ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0058 - mse: 0.0058 - mae: 0.0601 - val_loss: 0.0574 - val_mse: 0.0574 - val_mae: 0.2029 - lr: 1.0000e-05 - 58ms/epoch - 6ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0606 - val_loss: 0.0574 - val_mse: 0.0574 - val_mae: 0.2029 - lr: 1.0000e-05 - 66ms/epoch - 7ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0616 - val_loss: 0.0575 - val_mse: 0.0575 - val_mae: 0.2031 - lr: 1.0000e-05 - 53ms/epoch - 5ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0611 - val_loss: 0.0575 - val_mse: 0.0575 - val_mae: 0.2031 - lr: 1.0000e-05 - 71ms/epoch - 7ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0068 - mse: 0.0068 - mae: 0.0645 - val_loss: 0.0574 - val_mse: 0.0574 - val_mae: 0.2030 - lr: 1.0000e-05 - 58ms/epoch - 6ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0612 - val_loss: 0.0575 - val_mse: 0.0575 - val_mae: 0.2030 - lr: 1.0000e-05 - 57ms/epoch - 6ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0627 - val_loss: 0.0575 - val_mse: 0.0575 - val_mae: 0.2031 - lr: 1.0000e-05 - 53ms/epoch - 5ms/step
Epoch 56/500
Epoch 00056: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0067 - mse: 0.0067 - mae: 0.0642 - val_loss: 0.0575 - val_mse: 0.0575 - val_mae: 0.2031 - lr: 1.0000e-05 - 60ms/epoch - 6ms/step
Epoch 57/500
Epoch 00057: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0069 - mse: 0.0069 - mae: 0.0654 - val_loss: 0.0575 - val_mse: 0.0575 - val_mae: 0.2031 - lr: 1.0000e-05 - 56ms/epoch - 6ms/step
Epoch 58/500
Epoch 00058: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0067 - mse: 0.0067 - mae: 0.0643 - val_loss: 0.0574 - val_mse: 0.0574 - val_mae: 0.2030 - lr: 1.0000e-05 - 74ms/epoch - 7ms/step
Epoch 59/500
Epoch 00059: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0596 - val_loss: 0.0574 - val_mse: 0.0574 - val_mae: 0.2029 - lr: 1.0000e-05 - 64ms/epoch - 6ms/step
Epoch 60/500
Epoch 00060: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0645 - val_loss: 0.0573 - val_mse: 0.0573 - val_mae: 0.2027 - lr: 1.0000e-05 - 68ms/epoch - 7ms/step
Epoch 61/500
Epoch 00061: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0627 - val_loss: 0.0573 - val_mse: 0.0573 - val_mae: 0.2027 - lr: 1.0000e-05 - 110ms/epoch - 11ms/step
Epoch 62/500
Epoch 00062: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0065 - mse: 0.0065 - mae: 0.0625 - val_loss: 0.0573 - val_mse: 0.0573 - val_mae: 0.2027 - lr: 1.0000e-05 - 62ms/epoch - 6ms/step
Epoch 63/500
Epoch 00063: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0065 - mse: 0.0065 - mae: 0.0618 - val_loss: 0.0573 - val_mse: 0.0573 - val_mae: 0.2026 - lr: 1.0000e-05 - 76ms/epoch - 8ms/step
Epoch 64/500
Epoch 00064: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0616 - val_loss: 0.0572 - val_mse: 0.0572 - val_mae: 0.2026 - lr: 1.0000e-05 - 82ms/epoch - 8ms/step
Epoch 65/500
Epoch 00065: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0068 - mse: 0.0068 - mae: 0.0632 - val_loss: 0.0572 - val_mse: 0.0572 - val_mae: 0.2025 - lr: 1.0000e-05 - 58ms/epoch - 6ms/step
Epoch 66/500
Epoch 00066: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0606 - val_loss: 0.0572 - val_mse: 0.0572 - val_mae: 0.2024 - lr: 1.0000e-05 - 58ms/epoch - 6ms/step
Epoch 67/500
Epoch 00067: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0628 - val_loss: 0.0571 - val_mse: 0.0571 - val_mae: 0.2023 - lr: 1.0000e-05 - 66ms/epoch - 7ms/step
Epoch 68/500
Epoch 00068: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0620 - val_loss: 0.0571 - val_mse: 0.0571 - val_mae: 0.2023 - lr: 1.0000e-05 - 54ms/epoch - 5ms/step
Epoch 69/500
Epoch 00069: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0634 - val_loss: 0.0571 - val_mse: 0.0571 - val_mae: 0.2023 - lr: 1.0000e-05 - 59ms/epoch - 6ms/step
Epoch 70/500
Epoch 00070: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0071 - mse: 0.0071 - mae: 0.0654 - val_loss: 0.0571 - val_mse: 0.0571 - val_mae: 0.2022 - lr: 1.0000e-05 - 70ms/epoch - 7ms/step
Epoch 71/500
Epoch 00071: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0067 - mse: 0.0067 - mae: 0.0641 - val_loss: 0.0571 - val_mse: 0.0571 - val_mae: 0.2022 - lr: 1.0000e-05 - 79ms/epoch - 8ms/step
Epoch 72/500
Epoch 00072: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0610 - val_loss: 0.0571 - val_mse: 0.0571 - val_mae: 0.2022 - lr: 1.0000e-05 - 62ms/epoch - 6ms/step
Epoch 73/500
Epoch 00073: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0066 - mse: 0.0066 - mae: 0.0632 - val_loss: 0.0571 - val_mse: 0.0571 - val_mae: 0.2023 - lr: 1.0000e-05 - 61ms/epoch - 6ms/step
Epoch 74/500
Epoch 00074: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0620 - val_loss: 0.0571 - val_mse: 0.0571 - val_mae: 0.2023 - lr: 1.0000e-05 - 53ms/epoch - 5ms/step
Epoch 75/500
Epoch 00075: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0613 - val_loss: 0.0570 - val_mse: 0.0570 - val_mae: 0.2020 - lr: 1.0000e-05 - 67ms/epoch - 7ms/step
Epoch 76/500
Epoch 00076: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0065 - mse: 0.0065 - mae: 0.0625 - val_loss: 0.0570 - val_mse: 0.0570 - val_mae: 0.2020 - lr: 1.0000e-05 - 61ms/epoch - 6ms/step
Epoch 77/500
Epoch 00077: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0065 - mse: 0.0065 - mae: 0.0630 - val_loss: 0.0569 - val_mse: 0.0569 - val_mae: 0.2018 - lr: 1.0000e-05 - 75ms/epoch - 8ms/step
Epoch 78/500
Epoch 00078: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0609 - val_loss: 0.0569 - val_mse: 0.0569 - val_mae: 0.2017 - lr: 1.0000e-05 - 59ms/epoch - 6ms/step
Epoch 79/500
Epoch 00079: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0631 - val_loss: 0.0569 - val_mse: 0.0569 - val_mae: 0.2017 - lr: 1.0000e-05 - 61ms/epoch - 6ms/step
Epoch 80/500
Epoch 00080: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0065 - mse: 0.0065 - mae: 0.0630 - val_loss: 0.0568 - val_mse: 0.0568 - val_mae: 0.2017 - lr: 1.0000e-05 - 59ms/epoch - 6ms/step
Epoch 81/500
Epoch 00081: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0604 - val_loss: 0.0569 - val_mse: 0.0569 - val_mae: 0.2017 - lr: 1.0000e-05 - 68ms/epoch - 7ms/step
Epoch 82/500
Epoch 00082: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0610 - val_loss: 0.0570 - val_mse: 0.0570 - val_mae: 0.2019 - lr: 1.0000e-05 - 65ms/epoch - 7ms/step
Epoch 83/500
Epoch 00083: val_loss did not improve from 0.05221
10/10 - 0s - loss: 0.0064 - mse: 0.0064 - mae: 0.0623 - val_loss: 0.0570 - val_mse: 0.0570 - val_mae: 0.2020 - lr: 1.0000e-05 - 73ms/epoch - 7ms/step
Epoch 00083: early stopping
SMA
Prediction vs Close: 50.0% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 133.6684179910327
RMSE: 11.561505870388714
MAPE: 10.289389775089397
EMA
Prediction vs Close: 50.0% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 34.52931938659473
RMSE: 5.876165364129462
MAPE: 4.852473639818076
WMA
Prediction vs Close: 51.49% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 36.74468487644727
RMSE: 6.061739426637149
MAPE: 4.85767480758183
DEMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 87.45802496937279
RMSE: 9.351899538028238
MAPE: 8.239361009856534
KAMA
KAMA([input_arrays], [timeperiod=30])
Kaufman Adaptive Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
18
Working on KAMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16921.943, Time=10.66 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.592, Time=4.78 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-16797.275, Time=9.36 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.592, Time=6.82 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16996.465, Time=3.24 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=-16999.509, Time=3.22 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=-17171.315, Time=6.35 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=-16994.523, Time=4.16 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=-15518.026, Time=30.90 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 79.512 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood 8613.658
Date: Sun, 12 Dec 2021 AIC -17171.315
Time: 17:34:56 BIC -17039.972
Sample: 0 HQIC -17120.874
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -5.14e-10 7.6e-05 -6.76e-06 1.000 -0.000 0.000
x2 -5.041e-10 7.52e-05 -6.7e-06 1.000 -0.000 0.000
x3 -4.834e-10 7.38e-05 -6.55e-06 1.000 -0.000 0.000
x4 1.0000 7.46e-05 1.34e+04 0.000 1.000 1.000
x5 -4.462e-10 7.09e-05 -6.29e-06 1.000 -0.000 0.000
x6 -3.064e-09 0.000 -1.84e-05 1.000 -0.000 0.000
x7 -4.751e-10 7.35e-05 -6.46e-06 1.000 -0.000 0.000
x8 -4.628e-10 7.28e-05 -6.36e-06 1.000 -0.000 0.000
x9 -9.21e-11 9.37e-06 -9.83e-06 1.000 -1.84e-05 1.84e-05
x10 -2.165e-10 3.1e-05 -6.98e-06 1.000 -6.08e-05 6.08e-05
x11 -4.665e-10 7.28e-05 -6.41e-06 1.000 -0.000 0.000
x12 -4.62e-10 7.23e-05 -6.39e-06 1.000 -0.000 0.000
x13 -4.906e-10 7.43e-05 -6.6e-06 1.000 -0.000 0.000
x14 -3.985e-09 0.000 -1.87e-05 1.000 -0.000 0.000
x15 -4.897e-10 7.48e-05 -6.55e-06 1.000 -0.000 0.000
x16 -7.327e-10 9.24e-05 -7.93e-06 1.000 -0.000 0.000
x17 -4.173e-10 6.93e-05 -6.02e-06 1.000 -0.000 0.000
x18 -3.397e-10 6.02e-05 -5.64e-06 1.000 -0.000 0.000
x19 -6.012e-10 8.3e-05 -7.25e-06 1.000 -0.000 0.000
x20 -9.09e-10 0.000 -9.05e-06 1.000 -0.000 0.000
x21 -6.188e-09 0.000 -2.32e-05 1.000 -0.001 0.001
x22 -1.992e-09 0.000 -1.33e-05 1.000 -0.000 0.000
x23 -3.669e-09 0.000 -1.79e-05 1.000 -0.000 0.000
x24 -1.065e-09 0.000 -1.01e-05 1.000 -0.000 0.000
ar.L1 -1.2073 5.73e-10 -2.11e+09 0.000 -1.207 -1.207
ar.L2 -0.9083 5.93e-10 -1.53e+09 0.000 -0.908 -0.908
ar.L3 -0.4033 5.84e-10 -6.91e+08 0.000 -0.403 -0.403
sigma2 8.06e-11 6.94e-11 1.162 0.245 -5.54e-11 2.17e-10
===================================================================================
Ljung-Box (L1) (Q): 13.77 Jarque-Bera (JB): 2436796.68
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 4.07
Prob(H) (two-sided): 0.00 Kurtosis: 272.41
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.23e+28. Standard errors may be unstable.
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.04274, saving model to LSTM7.h5
45/45 - 2s - loss: 0.2395 - mse: 0.2395 - mae: 0.3786 - val_loss: 0.0427 - val_mse: 0.0427 - val_mae: 0.1818 - lr: 0.0010 - 2s/epoch - 54ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0676 - mse: 0.0676 - mae: 0.1991 - val_loss: 0.0669 - val_mse: 0.0669 - val_mae: 0.2104 - lr: 0.0010 - 206ms/epoch - 5ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0224 - mse: 0.0224 - mae: 0.1167 - val_loss: 0.1187 - val_mse: 0.1187 - val_mae: 0.2940 - lr: 0.0010 - 210ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0132 - mse: 0.0132 - mae: 0.0890 - val_loss: 0.0991 - val_mse: 0.0991 - val_mae: 0.2636 - lr: 0.0010 - 251ms/epoch - 6ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0124 - mse: 0.0124 - mae: 0.0873 - val_loss: 0.1059 - val_mse: 0.1059 - val_mae: 0.2774 - lr: 0.0010 - 222ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00006: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0101 - mse: 0.0101 - mae: 0.0811 - val_loss: 0.0999 - val_mse: 0.0999 - val_mae: 0.2688 - lr: 0.0010 - 234ms/epoch - 5ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0103 - mse: 0.0103 - mae: 0.0762 - val_loss: 0.0948 - val_mse: 0.0948 - val_mae: 0.2601 - lr: 1.0000e-04 - 200ms/epoch - 4ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0095 - mse: 0.0095 - mae: 0.0770 - val_loss: 0.0936 - val_mse: 0.0936 - val_mae: 0.2585 - lr: 1.0000e-04 - 210ms/epoch - 5ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0097 - mse: 0.0097 - mae: 0.0751 - val_loss: 0.0945 - val_mse: 0.0945 - val_mae: 0.2603 - lr: 1.0000e-04 - 222ms/epoch - 5ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0093 - mse: 0.0093 - mae: 0.0757 - val_loss: 0.0926 - val_mse: 0.0926 - val_mae: 0.2574 - lr: 1.0000e-04 - 208ms/epoch - 5ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00011: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0091 - mse: 0.0091 - mae: 0.0741 - val_loss: 0.0918 - val_mse: 0.0918 - val_mae: 0.2563 - lr: 1.0000e-04 - 202ms/epoch - 4ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0092 - mse: 0.0092 - mae: 0.0738 - val_loss: 0.0919 - val_mse: 0.0919 - val_mae: 0.2566 - lr: 1.0000e-05 - 203ms/epoch - 5ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0080 - mse: 0.0080 - mae: 0.0704 - val_loss: 0.0915 - val_mse: 0.0915 - val_mae: 0.2559 - lr: 1.0000e-05 - 220ms/epoch - 5ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0089 - mse: 0.0089 - mae: 0.0737 - val_loss: 0.0916 - val_mse: 0.0916 - val_mae: 0.2561 - lr: 1.0000e-05 - 206ms/epoch - 5ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0085 - mse: 0.0085 - mae: 0.0721 - val_loss: 0.0914 - val_mse: 0.0914 - val_mae: 0.2558 - lr: 1.0000e-05 - 236ms/epoch - 5ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00016: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0091 - mse: 0.0091 - mae: 0.0744 - val_loss: 0.0912 - val_mse: 0.0912 - val_mae: 0.2553 - lr: 1.0000e-05 - 226ms/epoch - 5ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0098 - mse: 0.0098 - mae: 0.0759 - val_loss: 0.0911 - val_mse: 0.0911 - val_mae: 0.2553 - lr: 1.0000e-05 - 241ms/epoch - 5ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0101 - mse: 0.0101 - mae: 0.0759 - val_loss: 0.0905 - val_mse: 0.0905 - val_mae: 0.2542 - lr: 1.0000e-05 - 230ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0092 - mse: 0.0092 - mae: 0.0743 - val_loss: 0.0906 - val_mse: 0.0906 - val_mae: 0.2545 - lr: 1.0000e-05 - 210ms/epoch - 5ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0088 - mse: 0.0088 - mae: 0.0733 - val_loss: 0.0906 - val_mse: 0.0906 - val_mae: 0.2544 - lr: 1.0000e-05 - 225ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0092 - mse: 0.0092 - mae: 0.0746 - val_loss: 0.0902 - val_mse: 0.0902 - val_mae: 0.2538 - lr: 1.0000e-05 - 201ms/epoch - 4ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0090 - mse: 0.0090 - mae: 0.0737 - val_loss: 0.0905 - val_mse: 0.0905 - val_mae: 0.2544 - lr: 1.0000e-05 - 188ms/epoch - 4ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0095 - mse: 0.0095 - mae: 0.0746 - val_loss: 0.0904 - val_mse: 0.0904 - val_mae: 0.2542 - lr: 1.0000e-05 - 229ms/epoch - 5ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0089 - mse: 0.0089 - mae: 0.0735 - val_loss: 0.0903 - val_mse: 0.0903 - val_mae: 0.2542 - lr: 1.0000e-05 - 187ms/epoch - 4ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0093 - mse: 0.0093 - mae: 0.0751 - val_loss: 0.0902 - val_mse: 0.0902 - val_mae: 0.2541 - lr: 1.0000e-05 - 199ms/epoch - 4ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0089 - mse: 0.0089 - mae: 0.0743 - val_loss: 0.0903 - val_mse: 0.0903 - val_mae: 0.2543 - lr: 1.0000e-05 - 189ms/epoch - 4ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0084 - mse: 0.0084 - mae: 0.0710 - val_loss: 0.0902 - val_mse: 0.0902 - val_mae: 0.2540 - lr: 1.0000e-05 - 193ms/epoch - 4ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0084 - mse: 0.0084 - mae: 0.0725 - val_loss: 0.0905 - val_mse: 0.0905 - val_mae: 0.2548 - lr: 1.0000e-05 - 198ms/epoch - 4ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0095 - mse: 0.0095 - mae: 0.0750 - val_loss: 0.0907 - val_mse: 0.0907 - val_mae: 0.2550 - lr: 1.0000e-05 - 210ms/epoch - 5ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0093 - mse: 0.0093 - mae: 0.0749 - val_loss: 0.0906 - val_mse: 0.0906 - val_mae: 0.2549 - lr: 1.0000e-05 - 198ms/epoch - 4ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0084 - mse: 0.0084 - mae: 0.0719 - val_loss: 0.0905 - val_mse: 0.0905 - val_mae: 0.2548 - lr: 1.0000e-05 - 210ms/epoch - 5ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0085 - mse: 0.0085 - mae: 0.0715 - val_loss: 0.0900 - val_mse: 0.0900 - val_mae: 0.2540 - lr: 1.0000e-05 - 185ms/epoch - 4ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0089 - mse: 0.0089 - mae: 0.0734 - val_loss: 0.0898 - val_mse: 0.0898 - val_mae: 0.2536 - lr: 1.0000e-05 - 205ms/epoch - 5ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0091 - mse: 0.0091 - mae: 0.0748 - val_loss: 0.0895 - val_mse: 0.0895 - val_mae: 0.2532 - lr: 1.0000e-05 - 174ms/epoch - 4ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0095 - mse: 0.0095 - mae: 0.0751 - val_loss: 0.0893 - val_mse: 0.0893 - val_mae: 0.2529 - lr: 1.0000e-05 - 203ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0082 - mse: 0.0082 - mae: 0.0697 - val_loss: 0.0889 - val_mse: 0.0889 - val_mae: 0.2522 - lr: 1.0000e-05 - 219ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0087 - mse: 0.0087 - mae: 0.0723 - val_loss: 0.0892 - val_mse: 0.0892 - val_mae: 0.2528 - lr: 1.0000e-05 - 189ms/epoch - 4ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0080 - mse: 0.0080 - mae: 0.0693 - val_loss: 0.0892 - val_mse: 0.0892 - val_mae: 0.2527 - lr: 1.0000e-05 - 250ms/epoch - 6ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0082 - mse: 0.0082 - mae: 0.0693 - val_loss: 0.0892 - val_mse: 0.0892 - val_mae: 0.2528 - lr: 1.0000e-05 - 239ms/epoch - 5ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0087 - mse: 0.0087 - mae: 0.0729 - val_loss: 0.0890 - val_mse: 0.0890 - val_mae: 0.2525 - lr: 1.0000e-05 - 214ms/epoch - 5ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0080 - mse: 0.0080 - mae: 0.0701 - val_loss: 0.0892 - val_mse: 0.0892 - val_mae: 0.2528 - lr: 1.0000e-05 - 202ms/epoch - 4ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0088 - mse: 0.0088 - mae: 0.0718 - val_loss: 0.0889 - val_mse: 0.0889 - val_mae: 0.2524 - lr: 1.0000e-05 - 214ms/epoch - 5ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0084 - mse: 0.0084 - mae: 0.0707 - val_loss: 0.0887 - val_mse: 0.0887 - val_mae: 0.2521 - lr: 1.0000e-05 - 203ms/epoch - 5ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0086 - mse: 0.0086 - mae: 0.0721 - val_loss: 0.0886 - val_mse: 0.0886 - val_mae: 0.2520 - lr: 1.0000e-05 - 218ms/epoch - 5ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0081 - mse: 0.0081 - mae: 0.0694 - val_loss: 0.0885 - val_mse: 0.0885 - val_mae: 0.2518 - lr: 1.0000e-05 - 182ms/epoch - 4ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0086 - mse: 0.0086 - mae: 0.0718 - val_loss: 0.0891 - val_mse: 0.0891 - val_mae: 0.2530 - lr: 1.0000e-05 - 240ms/epoch - 5ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0087 - mse: 0.0087 - mae: 0.0703 - val_loss: 0.0894 - val_mse: 0.0894 - val_mae: 0.2535 - lr: 1.0000e-05 - 204ms/epoch - 5ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0082 - mse: 0.0082 - mae: 0.0704 - val_loss: 0.0891 - val_mse: 0.0891 - val_mae: 0.2531 - lr: 1.0000e-05 - 226ms/epoch - 5ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0085 - mse: 0.0085 - mae: 0.0730 - val_loss: 0.0890 - val_mse: 0.0890 - val_mae: 0.2530 - lr: 1.0000e-05 - 224ms/epoch - 5ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0082 - mse: 0.0082 - mae: 0.0703 - val_loss: 0.0892 - val_mse: 0.0892 - val_mae: 0.2532 - lr: 1.0000e-05 - 173ms/epoch - 4ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.04274
45/45 - 0s - loss: 0.0083 - mse: 0.0083 - mae: 0.0697 - val_loss: 0.0892 - val_mse: 0.0892 - val_mae: 0.2533 - lr: 1.0000e-05 - 221ms/epoch - 5ms/step
Epoch 00051: early stopping
SMA
Prediction vs Close: 50.0% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 133.6684179910327
RMSE: 11.561505870388714
MAPE: 10.289389775089397
EMA
Prediction vs Close: 50.0% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 34.52931938659473
RMSE: 5.876165364129462
MAPE: 4.852473639818076
WMA
Prediction vs Close: 51.49% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 36.74468487644727
RMSE: 6.061739426637149
MAPE: 4.85767480758183
DEMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 87.45802496937279
RMSE: 9.351899538028238
MAPE: 8.239361009856534
KAMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 55.655998887145394
RMSE: 7.460294825752223
MAPE: 6.325008398714769
MIDPOINT
MIDPOINT([input_arrays], [timeperiod=14])
MidPoint over period (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 14
Outputs:
real
14
Working on MIDPOINT predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.768, Time=3.42 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.591, Time=4.75 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-15581.065, Time=8.84 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.591, Time=7.41 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16536.628, Time=9.30 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-13971.493, Time=10.19 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17226.044, Time=21.94 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-14562.591, Time=9.57 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-16754.945, Time=20.60 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-15001.855, Time=20.53 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 116.545 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8640.022
Date: Sun, 12 Dec 2021 AIC -17226.044
Time: 17:38:50 BIC -17099.391
Sample: 0 HQIC -17177.404
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -5.031e-09 1.06e-05 -0.000 1.000 -2.08e-05 2.08e-05
x2 -4.99e-09 8.12e-06 -0.001 1.000 -1.59e-05 1.59e-05
x3 -5.114e-09 1.38e-05 -0.000 1.000 -2.7e-05 2.7e-05
x4 1.0000 8.91e-06 1.12e+05 0.000 1.000 1.000
x5 -4.55e-09 8.2e-06 -0.001 1.000 -1.61e-05 1.61e-05
x6 -9.992e-08 0.001 -0.000 1.000 -0.002 0.002
x7 -4.607e-09 1.97e-05 -0.000 1.000 -3.86e-05 3.86e-05
x8 -4.591e-09 1.77e-05 -0.000 1.000 -3.48e-05 3.48e-05
x9 -2.538e-09 1.13e-05 -0.000 1.000 -2.21e-05 2.21e-05
x10 -4.315e-09 6.08e-06 -0.001 0.999 -1.19e-05 1.19e-05
x11 -4.545e-09 1.62e-05 -0.000 1.000 -3.18e-05 3.18e-05
x12 -4.701e-09 1.97e-05 -0.000 1.000 -3.87e-05 3.87e-05
x13 -4.823e-09 1.18e-05 -0.000 1.000 -2.3e-05 2.3e-05
x14 -4.08e-08 4.99e-05 -0.001 0.999 -9.79e-05 9.78e-05
x15 -5.557e-09 2.03e-05 -0.000 1.000 -3.99e-05 3.99e-05
x16 -3.541e-09 1.3e-05 -0.000 1.000 -2.55e-05 2.55e-05
x17 -3.463e-09 1.51e-05 -0.000 1.000 -2.97e-05 2.97e-05
x18 -1.534e-08 4e-05 -0.000 1.000 -7.85e-05 7.85e-05
x19 -6.118e-09 2.07e-05 -0.000 1.000 -4.05e-05 4.05e-05
x20 -1.581e-08 3.38e-05 -0.000 1.000 -6.62e-05 6.61e-05
x21 -5.505e-08 5.6e-05 -0.001 0.999 -0.000 0.000
x22 -2.936e-08 4.55e-05 -0.001 0.999 -8.92e-05 8.92e-05
x23 -3.882e-08 4.89e-05 -0.001 0.999 -9.58e-05 9.57e-05
x24 -2.099e-08 4.87e-05 -0.000 1.000 -9.54e-05 9.54e-05
ma.L1 -1.3900 1.23e-07 -1.13e+07 0.000 -1.390 -1.390
ma.L2 0.4044 1.43e-07 2.82e+06 0.000 0.404 0.404
sigma2 7.525e-11 7.22e-11 1.042 0.297 -6.63e-11 2.17e-10
===================================================================================
Ljung-Box (L1) (Q): 55.84 Jarque-Bera (JB): 1335305.59
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.09 Skew: 5.74
Prob(H) (two-sided): 0.00 Kurtosis: 202.19
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.77e+23. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.42203, saving model to LSTM7.h5
58/58 - 3s - loss: 0.3093 - mse: 0.3093 - mae: 0.3561 - val_loss: 0.4220 - val_mse: 0.4220 - val_mae: 0.6374 - lr: 0.0010 - 3s/epoch - 51ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.42203 to 0.25761, saving model to LSTM7.h5
58/58 - 0s - loss: 0.0252 - mse: 0.0252 - mae: 0.1265 - val_loss: 0.2576 - val_mse: 0.2576 - val_mae: 0.4938 - lr: 0.0010 - 305ms/epoch - 5ms/step
Epoch 3/500
Epoch 00003: val_loss improved from 0.25761 to 0.18104, saving model to LSTM7.h5
58/58 - 0s - loss: 0.0194 - mse: 0.0194 - mae: 0.1105 - val_loss: 0.1810 - val_mse: 0.1810 - val_mae: 0.4104 - lr: 0.0010 - 281ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss improved from 0.18104 to 0.13620, saving model to LSTM7.h5
58/58 - 0s - loss: 0.0157 - mse: 0.0157 - mae: 0.1005 - val_loss: 0.1362 - val_mse: 0.1362 - val_mae: 0.3531 - lr: 0.0010 - 286ms/epoch - 5ms/step
Epoch 5/500
Epoch 00005: val_loss improved from 0.13620 to 0.10533, saving model to LSTM7.h5
58/58 - 0s - loss: 0.0120 - mse: 0.0120 - mae: 0.0861 - val_loss: 0.1053 - val_mse: 0.1053 - val_mae: 0.3076 - lr: 0.0010 - 266ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: val_loss improved from 0.10533 to 0.09835, saving model to LSTM7.h5
58/58 - 0s - loss: 0.0124 - mse: 0.0124 - mae: 0.0873 - val_loss: 0.0983 - val_mse: 0.0983 - val_mae: 0.2965 - lr: 0.0010 - 250ms/epoch - 4ms/step
Epoch 7/500
Epoch 00007: val_loss improved from 0.09835 to 0.07233, saving model to LSTM7.h5
58/58 - 0s - loss: 0.0104 - mse: 0.0104 - mae: 0.0808 - val_loss: 0.0723 - val_mse: 0.0723 - val_mae: 0.2497 - lr: 0.0010 - 277ms/epoch - 5ms/step
Epoch 8/500
Epoch 00008: val_loss improved from 0.07233 to 0.07117, saving model to LSTM7.h5
58/58 - 0s - loss: 0.0096 - mse: 0.0096 - mae: 0.0778 - val_loss: 0.0712 - val_mse: 0.0712 - val_mae: 0.2472 - lr: 0.0010 - 264ms/epoch - 5ms/step
Epoch 9/500
Epoch 00009: val_loss improved from 0.07117 to 0.04473, saving model to LSTM7.h5
58/58 - 0s - loss: 0.0087 - mse: 0.0087 - mae: 0.0739 - val_loss: 0.0447 - val_mse: 0.0447 - val_mae: 0.1884 - lr: 0.0010 - 287ms/epoch - 5ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.04473
58/58 - 0s - loss: 0.0099 - mse: 0.0099 - mae: 0.0787 - val_loss: 0.0562 - val_mse: 0.0562 - val_mae: 0.2155 - lr: 0.0010 - 255ms/epoch - 4ms/step
Epoch 11/500
Epoch 00011: val_loss improved from 0.04473 to 0.04300, saving model to LSTM7.h5
58/58 - 0s - loss: 0.0073 - mse: 0.0073 - mae: 0.0682 - val_loss: 0.0430 - val_mse: 0.0430 - val_mae: 0.1846 - lr: 0.0010 - 255ms/epoch - 4ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.04300
58/58 - 0s - loss: 0.0070 - mse: 0.0070 - mae: 0.0679 - val_loss: 0.0515 - val_mse: 0.0515 - val_mae: 0.2046 - lr: 0.0010 - 257ms/epoch - 4ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.04300
58/58 - 0s - loss: 0.0070 - mse: 0.0070 - mae: 0.0663 - val_loss: 0.0443 - val_mse: 0.0443 - val_mae: 0.1869 - lr: 0.0010 - 268ms/epoch - 5ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.04300
58/58 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0614 - val_loss: 0.0594 - val_mse: 0.0594 - val_mae: 0.2221 - lr: 0.0010 - 281ms/epoch - 5ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.04300
58/58 - 0s - loss: 0.0063 - mse: 0.0063 - mae: 0.0627 - val_loss: 0.0463 - val_mse: 0.0463 - val_mae: 0.1911 - lr: 0.0010 - 259ms/epoch - 4ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00016: val_loss did not improve from 0.04300
58/58 - 0s - loss: 0.0059 - mse: 0.0059 - mae: 0.0619 - val_loss: 0.0496 - val_mse: 0.0496 - val_mae: 0.1986 - lr: 0.0010 - 262ms/epoch - 5ms/step
Epoch 17/500
Epoch 00017: val_loss improved from 0.04300 to 0.03942, saving model to LSTM7.h5
58/58 - 0s - loss: 0.0075 - mse: 0.0075 - mae: 0.0679 - val_loss: 0.0394 - val_mse: 0.0394 - val_mae: 0.1730 - lr: 1.0000e-04 - 284ms/epoch - 5ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0626 - val_loss: 0.0411 - val_mse: 0.0411 - val_mae: 0.1774 - lr: 1.0000e-04 - 255ms/epoch - 4ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0563 - val_loss: 0.0432 - val_mse: 0.0432 - val_mae: 0.1829 - lr: 1.0000e-04 - 254ms/epoch - 4ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0048 - mse: 0.0048 - mae: 0.0546 - val_loss: 0.0439 - val_mse: 0.0439 - val_mae: 0.1847 - lr: 1.0000e-04 - 275ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0539 - val_loss: 0.0449 - val_mse: 0.0449 - val_mae: 0.1872 - lr: 1.0000e-04 - 270ms/epoch - 5ms/step
Epoch 22/500
Epoch 00022: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00022: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0043 - mse: 0.0043 - mae: 0.0514 - val_loss: 0.0454 - val_mse: 0.0454 - val_mae: 0.1884 - lr: 1.0000e-04 - 270ms/epoch - 5ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0040 - mse: 0.0040 - mae: 0.0486 - val_loss: 0.0454 - val_mse: 0.0454 - val_mae: 0.1883 - lr: 1.0000e-05 - 308ms/epoch - 5ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0507 - val_loss: 0.0452 - val_mse: 0.0452 - val_mae: 0.1879 - lr: 1.0000e-05 - 287ms/epoch - 5ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0041 - mse: 0.0041 - mae: 0.0498 - val_loss: 0.0452 - val_mse: 0.0452 - val_mae: 0.1880 - lr: 1.0000e-05 - 226ms/epoch - 4ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0511 - val_loss: 0.0452 - val_mse: 0.0452 - val_mae: 0.1879 - lr: 1.0000e-05 - 266ms/epoch - 5ms/step
Epoch 27/500
Epoch 00027: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00027: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0515 - val_loss: 0.0453 - val_mse: 0.0453 - val_mae: 0.1882 - lr: 1.0000e-05 - 254ms/epoch - 4ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0043 - mse: 0.0043 - mae: 0.0512 - val_loss: 0.0453 - val_mse: 0.0453 - val_mae: 0.1882 - lr: 1.0000e-05 - 266ms/epoch - 5ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0041 - mse: 0.0041 - mae: 0.0497 - val_loss: 0.0452 - val_mse: 0.0452 - val_mae: 0.1879 - lr: 1.0000e-05 - 247ms/epoch - 4ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0508 - val_loss: 0.0452 - val_mse: 0.0452 - val_mae: 0.1878 - lr: 1.0000e-05 - 283ms/epoch - 5ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0510 - val_loss: 0.0451 - val_mse: 0.0451 - val_mae: 0.1877 - lr: 1.0000e-05 - 251ms/epoch - 4ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0043 - mse: 0.0043 - mae: 0.0517 - val_loss: 0.0450 - val_mse: 0.0450 - val_mae: 0.1874 - lr: 1.0000e-05 - 239ms/epoch - 4ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0534 - val_loss: 0.0449 - val_mse: 0.0449 - val_mae: 0.1872 - lr: 1.0000e-05 - 265ms/epoch - 5ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0040 - mse: 0.0040 - mae: 0.0490 - val_loss: 0.0449 - val_mse: 0.0449 - val_mae: 0.1872 - lr: 1.0000e-05 - 284ms/epoch - 5ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0490 - val_loss: 0.0448 - val_mse: 0.0448 - val_mae: 0.1870 - lr: 1.0000e-05 - 265ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0036 - mse: 0.0036 - mae: 0.0476 - val_loss: 0.0451 - val_mse: 0.0451 - val_mae: 0.1878 - lr: 1.0000e-05 - 249ms/epoch - 4ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0507 - val_loss: 0.0452 - val_mse: 0.0452 - val_mae: 0.1878 - lr: 1.0000e-05 - 258ms/epoch - 4ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0041 - mse: 0.0041 - mae: 0.0511 - val_loss: 0.0452 - val_mse: 0.0452 - val_mae: 0.1880 - lr: 1.0000e-05 - 266ms/epoch - 5ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0045 - mse: 0.0045 - mae: 0.0530 - val_loss: 0.0452 - val_mse: 0.0452 - val_mae: 0.1880 - lr: 1.0000e-05 - 256ms/epoch - 4ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0040 - mse: 0.0040 - mae: 0.0500 - val_loss: 0.0451 - val_mse: 0.0451 - val_mae: 0.1876 - lr: 1.0000e-05 - 281ms/epoch - 5ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0039 - mse: 0.0039 - mae: 0.0481 - val_loss: 0.0451 - val_mse: 0.0451 - val_mae: 0.1876 - lr: 1.0000e-05 - 301ms/epoch - 5ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0041 - mse: 0.0041 - mae: 0.0492 - val_loss: 0.0447 - val_mse: 0.0447 - val_mae: 0.1867 - lr: 1.0000e-05 - 294ms/epoch - 5ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0038 - mse: 0.0038 - mae: 0.0481 - val_loss: 0.0450 - val_mse: 0.0450 - val_mae: 0.1875 - lr: 1.0000e-05 - 293ms/epoch - 5ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0039 - mse: 0.0039 - mae: 0.0481 - val_loss: 0.0451 - val_mse: 0.0451 - val_mae: 0.1878 - lr: 1.0000e-05 - 315ms/epoch - 5ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0506 - val_loss: 0.0452 - val_mse: 0.0452 - val_mae: 0.1880 - lr: 1.0000e-05 - 275ms/epoch - 5ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0040 - mse: 0.0040 - mae: 0.0502 - val_loss: 0.0452 - val_mse: 0.0452 - val_mae: 0.1880 - lr: 1.0000e-05 - 265ms/epoch - 5ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0039 - mse: 0.0039 - mae: 0.0491 - val_loss: 0.0453 - val_mse: 0.0453 - val_mae: 0.1882 - lr: 1.0000e-05 - 262ms/epoch - 5ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0499 - val_loss: 0.0453 - val_mse: 0.0453 - val_mae: 0.1881 - lr: 1.0000e-05 - 292ms/epoch - 5ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0041 - mse: 0.0041 - mae: 0.0505 - val_loss: 0.0452 - val_mse: 0.0452 - val_mae: 0.1879 - lr: 1.0000e-05 - 251ms/epoch - 4ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0038 - mse: 0.0038 - mae: 0.0480 - val_loss: 0.0456 - val_mse: 0.0456 - val_mae: 0.1887 - lr: 1.0000e-05 - 250ms/epoch - 4ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0040 - mse: 0.0040 - mae: 0.0485 - val_loss: 0.0455 - val_mse: 0.0455 - val_mae: 0.1885 - lr: 1.0000e-05 - 263ms/epoch - 5ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0040 - mse: 0.0040 - mae: 0.0488 - val_loss: 0.0453 - val_mse: 0.0453 - val_mae: 0.1881 - lr: 1.0000e-05 - 277ms/epoch - 5ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0038 - mse: 0.0038 - mae: 0.0477 - val_loss: 0.0453 - val_mse: 0.0453 - val_mae: 0.1881 - lr: 1.0000e-05 - 266ms/epoch - 5ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0036 - mse: 0.0036 - mae: 0.0473 - val_loss: 0.0456 - val_mse: 0.0456 - val_mae: 0.1887 - lr: 1.0000e-05 - 231ms/epoch - 4ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0034 - mse: 0.0034 - mae: 0.0457 - val_loss: 0.0455 - val_mse: 0.0455 - val_mae: 0.1885 - lr: 1.0000e-05 - 277ms/epoch - 5ms/step
Epoch 56/500
Epoch 00056: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0041 - mse: 0.0041 - mae: 0.0488 - val_loss: 0.0455 - val_mse: 0.0455 - val_mae: 0.1886 - lr: 1.0000e-05 - 267ms/epoch - 5ms/step
Epoch 57/500
Epoch 00057: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0039 - mse: 0.0039 - mae: 0.0486 - val_loss: 0.0456 - val_mse: 0.0456 - val_mae: 0.1887 - lr: 1.0000e-05 - 239ms/epoch - 4ms/step
Epoch 58/500
Epoch 00058: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0040 - mse: 0.0040 - mae: 0.0490 - val_loss: 0.0454 - val_mse: 0.0454 - val_mae: 0.1882 - lr: 1.0000e-05 - 239ms/epoch - 4ms/step
Epoch 59/500
Epoch 00059: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0037 - mse: 0.0037 - mae: 0.0470 - val_loss: 0.0456 - val_mse: 0.0456 - val_mae: 0.1888 - lr: 1.0000e-05 - 262ms/epoch - 5ms/step
Epoch 60/500
Epoch 00060: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0040 - mse: 0.0040 - mae: 0.0490 - val_loss: 0.0457 - val_mse: 0.0457 - val_mae: 0.1889 - lr: 1.0000e-05 - 270ms/epoch - 5ms/step
Epoch 61/500
Epoch 00061: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0036 - mse: 0.0036 - mae: 0.0459 - val_loss: 0.0457 - val_mse: 0.0457 - val_mae: 0.1888 - lr: 1.0000e-05 - 271ms/epoch - 5ms/step
Epoch 62/500
Epoch 00062: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0038 - mse: 0.0038 - mae: 0.0477 - val_loss: 0.0457 - val_mse: 0.0457 - val_mae: 0.1888 - lr: 1.0000e-05 - 268ms/epoch - 5ms/step
Epoch 63/500
Epoch 00063: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0045 - mse: 0.0045 - mae: 0.0519 - val_loss: 0.0457 - val_mse: 0.0457 - val_mae: 0.1888 - lr: 1.0000e-05 - 258ms/epoch - 4ms/step
Epoch 64/500
Epoch 00064: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0037 - mse: 0.0037 - mae: 0.0466 - val_loss: 0.0459 - val_mse: 0.0459 - val_mae: 0.1893 - lr: 1.0000e-05 - 294ms/epoch - 5ms/step
Epoch 65/500
Epoch 00065: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0041 - mse: 0.0041 - mae: 0.0500 - val_loss: 0.0459 - val_mse: 0.0459 - val_mae: 0.1894 - lr: 1.0000e-05 - 258ms/epoch - 4ms/step
Epoch 66/500
Epoch 00066: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0039 - mse: 0.0039 - mae: 0.0488 - val_loss: 0.0461 - val_mse: 0.0461 - val_mae: 0.1898 - lr: 1.0000e-05 - 261ms/epoch - 4ms/step
Epoch 67/500
Epoch 00067: val_loss did not improve from 0.03942
58/58 - 0s - loss: 0.0038 - mse: 0.0038 - mae: 0.0481 - val_loss: 0.0459 - val_mse: 0.0459 - val_mae: 0.1893 - lr: 1.0000e-05 - 263ms/epoch - 5ms/step
Epoch 00067: early stopping
SMA
Prediction vs Close: 50.0% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 133.6684179910327
RMSE: 11.561505870388714
MAPE: 10.289389775089397
EMA
Prediction vs Close: 50.0% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 34.52931938659473
RMSE: 5.876165364129462
MAPE: 4.852473639818076
WMA
Prediction vs Close: 51.49% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 36.74468487644727
RMSE: 6.061739426637149
MAPE: 4.85767480758183
DEMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 87.45802496937279
RMSE: 9.351899538028238
MAPE: 8.239361009856534
KAMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 55.655998887145394
RMSE: 7.460294825752223
MAPE: 6.325008398714769
MIDPOINT
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 43.66% Accuracy
MSE: 28.58496814832563
RMSE: 5.346491199686541
MAPE: 4.412567377496203
T3
T3([input_arrays], [timeperiod=5], [vfactor=0.7])
Triple Exponential Moving Average (T3) (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 5
vfactor: 0.7
Outputs:
real
19
Working on T3 predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-17000.569, Time=3.09 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-15576.554, Time=5.76 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-16078.305, Time=8.09 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-15574.554, Time=9.35 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16998.627, Time=3.13 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-16429.916, Time=12.60 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17000.664, Time=3.17 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-15700.026, Time=12.26 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-15704.282, Time=15.13 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-16998.664, Time=3.57 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 76.160 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8527.332
Date: Sun, 12 Dec 2021 AIC -17000.664
Time: 17:44:29 BIC -16874.011
Sample: 0 HQIC -16952.024
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 8.378e-14 2.16e-06 3.89e-08 1.000 -4.23e-06 4.23e-06
x2 7.457e-14 2.15e-06 3.47e-08 1.000 -4.22e-06 4.22e-06
x3 2.279e-14 2.16e-06 1.05e-08 1.000 -4.24e-06 4.24e-06
x4 1.0000 2.16e-06 4.63e+05 0.000 1.000 1.000
x5 1.211e-12 2.07e-06 5.86e-07 1.000 -4.05e-06 4.05e-06
x6 3.146e-15 2.67e-06 1.18e-09 1.000 -5.23e-06 5.23e-06
x7 1.593e-13 2.15e-06 7.41e-08 1.000 -4.21e-06 4.21e-06
x8 -0.0001 2.1e-06 -48.778 0.000 -0.000 -9.82e-05
x9 5.141e-14 6.35e-07 8.1e-08 1.000 -1.24e-06 1.24e-06
x10 -6.174e-05 1.34e-06 -45.995 0.000 -6.44e-05 -5.91e-05
x11 0.0003 2.15e-06 148.354 0.000 0.000 0.000
x12 -0.0002 2.02e-06 -93.730 0.000 -0.000 -0.000
x13 1.967e-14 2.16e-06 9.1e-09 1.000 -4.23e-06 4.23e-06
x14 -1.297e-14 5.65e-06 -2.29e-09 1.000 -1.11e-05 1.11e-05
x15 -3.18e-12 1.82e-06 -1.75e-06 1.000 -3.57e-06 3.57e-06
x16 -1.426e-12 4.51e-06 -3.16e-07 1.000 -8.84e-06 8.84e-06
x17 7.474e-13 2.37e-06 3.16e-07 1.000 -4.64e-06 4.64e-06
x18 -2.92e-13 2.9e-06 -1.01e-07 1.000 -5.68e-06 5.68e-06
x19 -4.211e-14 1.89e-06 -2.22e-08 1.000 -3.71e-06 3.71e-06
x20 -1.515e-13 1.2e-06 -1.26e-07 1.000 -2.36e-06 2.36e-06
x21 6.555e-13 6.37e-06 1.03e-07 1.000 -1.25e-05 1.25e-05
x22 1.212e-14 6.19e-06 1.96e-09 1.000 -1.21e-05 1.21e-05
x23 -3.877e-13 3.76e-06 -1.03e-07 1.000 -7.38e-06 7.38e-06
x24 8.127e-15 4.01e-06 2.03e-09 1.000 -7.86e-06 7.86e-06
ma.L1 -1.3370 3.84e-12 -3.48e+11 0.000 -1.337 -1.337
ma.L2 0.4289 1.65e-12 2.6e+11 0.000 0.429 0.429
sigma2 1e-10 6.99e-11 1.430 0.153 -3.71e-11 2.37e-10
===================================================================================
Ljung-Box (L1) (Q): 4.57 Jarque-Bera (JB): 3228712.87
Prob(Q): 0.03 Prob(JB): 0.00
Heteroskedasticity (H): 0.12 Skew: -9.87
Prob(H) (two-sided): 0.00 Kurtosis: 312.63
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.57e+30. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.03552, saving model to LSTM7.h5
43/43 - 2s - loss: 0.3092 - mse: 0.3092 - mae: 0.3824 - val_loss: 0.0355 - val_mse: 0.0355 - val_mae: 0.1573 - lr: 0.0010 - 2s/epoch - 56ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.03552 to 0.03289, saving model to LSTM7.h5
43/43 - 0s - loss: 0.0320 - mse: 0.0320 - mae: 0.1413 - val_loss: 0.0329 - val_mse: 0.0329 - val_mae: 0.1486 - lr: 0.0010 - 227ms/epoch - 5ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.03289
43/43 - 0s - loss: 0.0216 - mse: 0.0216 - mae: 0.1169 - val_loss: 0.0346 - val_mse: 0.0346 - val_mae: 0.1516 - lr: 0.0010 - 194ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.03289
43/43 - 0s - loss: 0.0166 - mse: 0.0166 - mae: 0.1031 - val_loss: 0.0329 - val_mse: 0.0329 - val_mae: 0.1479 - lr: 0.0010 - 196ms/epoch - 5ms/step
Epoch 5/500
Epoch 00005: val_loss improved from 0.03289 to 0.02738, saving model to LSTM7.h5
43/43 - 0s - loss: 0.0137 - mse: 0.0137 - mae: 0.0924 - val_loss: 0.0274 - val_mse: 0.0274 - val_mae: 0.1340 - lr: 0.0010 - 212ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: val_loss improved from 0.02738 to 0.02701, saving model to LSTM7.h5
43/43 - 0s - loss: 0.0112 - mse: 0.0112 - mae: 0.0823 - val_loss: 0.0270 - val_mse: 0.0270 - val_mae: 0.1335 - lr: 0.0010 - 194ms/epoch - 5ms/step
Epoch 7/500
Epoch 00007: val_loss improved from 0.02701 to 0.02053, saving model to LSTM7.h5
43/43 - 0s - loss: 0.0105 - mse: 0.0105 - mae: 0.0819 - val_loss: 0.0205 - val_mse: 0.0205 - val_mae: 0.1143 - lr: 0.0010 - 188ms/epoch - 4ms/step
Epoch 8/500
Epoch 00008: val_loss improved from 0.02053 to 0.01771, saving model to LSTM7.h5
43/43 - 0s - loss: 0.0102 - mse: 0.0102 - mae: 0.0802 - val_loss: 0.0177 - val_mse: 0.0177 - val_mae: 0.1057 - lr: 0.0010 - 210ms/epoch - 5ms/step
Epoch 9/500
Epoch 00009: val_loss improved from 0.01771 to 0.01332, saving model to LSTM7.h5
43/43 - 0s - loss: 0.0088 - mse: 0.0088 - mae: 0.0745 - val_loss: 0.0133 - val_mse: 0.0133 - val_mae: 0.0930 - lr: 0.0010 - 223ms/epoch - 5ms/step
Epoch 10/500
Epoch 00010: val_loss improved from 0.01332 to 0.01268, saving model to LSTM7.h5
43/43 - 0s - loss: 0.0086 - mse: 0.0086 - mae: 0.0725 - val_loss: 0.0127 - val_mse: 0.0127 - val_mae: 0.0895 - lr: 0.0010 - 200ms/epoch - 5ms/step
Epoch 11/500
Epoch 00011: val_loss improved from 0.01268 to 0.01164, saving model to LSTM7.h5
43/43 - 0s - loss: 0.0073 - mse: 0.0073 - mae: 0.0678 - val_loss: 0.0116 - val_mse: 0.0116 - val_mae: 0.0861 - lr: 0.0010 - 234ms/epoch - 5ms/step
Epoch 12/500
Epoch 00012: val_loss improved from 0.01164 to 0.01039, saving model to LSTM7.h5
43/43 - 0s - loss: 0.0068 - mse: 0.0068 - mae: 0.0654 - val_loss: 0.0104 - val_mse: 0.0104 - val_mae: 0.0826 - lr: 0.0010 - 232ms/epoch - 5ms/step
Epoch 13/500
Epoch 00013: val_loss improved from 0.01039 to 0.00954, saving model to LSTM7.h5
43/43 - 0s - loss: 0.0073 - mse: 0.0073 - mae: 0.0676 - val_loss: 0.0095 - val_mse: 0.0095 - val_mae: 0.0808 - lr: 0.0010 - 230ms/epoch - 5ms/step
Epoch 14/500
Epoch 00014: val_loss improved from 0.00954 to 0.00885, saving model to LSTM7.h5
43/43 - 0s - loss: 0.0062 - mse: 0.0062 - mae: 0.0614 - val_loss: 0.0088 - val_mse: 0.0088 - val_mae: 0.0771 - lr: 0.0010 - 239ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: val_loss improved from 0.00885 to 0.00870, saving model to LSTM7.h5
43/43 - 0s - loss: 0.0063 - mse: 0.0063 - mae: 0.0624 - val_loss: 0.0087 - val_mse: 0.0087 - val_mae: 0.0785 - lr: 0.0010 - 196ms/epoch - 5ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.00870
43/43 - 0s - loss: 0.0060 - mse: 0.0060 - mae: 0.0601 - val_loss: 0.0099 - val_mse: 0.0099 - val_mae: 0.0849 - lr: 0.0010 - 174ms/epoch - 4ms/step
Epoch 17/500
Epoch 00017: val_loss improved from 0.00870 to 0.00834, saving model to LSTM7.h5
43/43 - 0s - loss: 0.0055 - mse: 0.0055 - mae: 0.0577 - val_loss: 0.0083 - val_mse: 0.0083 - val_mae: 0.0776 - lr: 0.0010 - 211ms/epoch - 5ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0050 - mse: 0.0050 - mae: 0.0567 - val_loss: 0.0102 - val_mse: 0.0102 - val_mae: 0.0856 - lr: 0.0010 - 194ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0539 - val_loss: 0.0104 - val_mse: 0.0104 - val_mae: 0.0859 - lr: 0.0010 - 181ms/epoch - 4ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0048 - mse: 0.0048 - mae: 0.0538 - val_loss: 0.0114 - val_mse: 0.0114 - val_mae: 0.0898 - lr: 0.0010 - 221ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0052 - mse: 0.0052 - mae: 0.0560 - val_loss: 0.0155 - val_mse: 0.0155 - val_mae: 0.1046 - lr: 0.0010 - 184ms/epoch - 4ms/step
Epoch 22/500
Epoch 00022: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00022: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0045 - mse: 0.0045 - mae: 0.0535 - val_loss: 0.0100 - val_mse: 0.0100 - val_mae: 0.0843 - lr: 0.0010 - 221ms/epoch - 5ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0045 - mse: 0.0045 - mae: 0.0520 - val_loss: 0.0098 - val_mse: 0.0098 - val_mae: 0.0837 - lr: 1.0000e-04 - 201ms/epoch - 5ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0545 - val_loss: 0.0099 - val_mse: 0.0099 - val_mae: 0.0844 - lr: 1.0000e-04 - 219ms/epoch - 5ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0040 - mse: 0.0040 - mae: 0.0499 - val_loss: 0.0104 - val_mse: 0.0104 - val_mae: 0.0864 - lr: 1.0000e-04 - 196ms/epoch - 5ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0514 - val_loss: 0.0106 - val_mse: 0.0106 - val_mae: 0.0873 - lr: 1.0000e-04 - 220ms/epoch - 5ms/step
Epoch 27/500
Epoch 00027: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00027: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0038 - mse: 0.0038 - mae: 0.0498 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0891 - lr: 1.0000e-04 - 184ms/epoch - 4ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0037 - mse: 0.0037 - mae: 0.0478 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0890 - lr: 1.0000e-05 - 199ms/epoch - 5ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0035 - mse: 0.0035 - mae: 0.0462 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0890 - lr: 1.0000e-05 - 176ms/epoch - 4ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0037 - mse: 0.0037 - mae: 0.0472 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0892 - lr: 1.0000e-05 - 197ms/epoch - 5ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0036 - mse: 0.0036 - mae: 0.0472 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0892 - lr: 1.0000e-05 - 221ms/epoch - 5ms/step
Epoch 32/500
Epoch 00032: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00032: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0039 - mse: 0.0039 - mae: 0.0486 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0891 - lr: 1.0000e-05 - 221ms/epoch - 5ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0034 - mse: 0.0034 - mae: 0.0450 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0891 - lr: 1.0000e-05 - 175ms/epoch - 4ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0036 - mse: 0.0036 - mae: 0.0472 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0890 - lr: 1.0000e-05 - 219ms/epoch - 5ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0035 - mse: 0.0035 - mae: 0.0462 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0891 - lr: 1.0000e-05 - 184ms/epoch - 4ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0035 - mse: 0.0035 - mae: 0.0466 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0894 - lr: 1.0000e-05 - 204ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0041 - mse: 0.0041 - mae: 0.0481 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0893 - lr: 1.0000e-05 - 232ms/epoch - 5ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0035 - mse: 0.0035 - mae: 0.0469 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0894 - lr: 1.0000e-05 - 199ms/epoch - 5ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0039 - mse: 0.0039 - mae: 0.0486 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0896 - lr: 1.0000e-05 - 218ms/epoch - 5ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0037 - mse: 0.0037 - mae: 0.0473 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0896 - lr: 1.0000e-05 - 185ms/epoch - 4ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0035 - mse: 0.0035 - mae: 0.0464 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0896 - lr: 1.0000e-05 - 227ms/epoch - 5ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0038 - mse: 0.0038 - mae: 0.0470 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0896 - lr: 1.0000e-05 - 186ms/epoch - 4ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0036 - mse: 0.0036 - mae: 0.0475 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0895 - lr: 1.0000e-05 - 222ms/epoch - 5ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0036 - mse: 0.0036 - mae: 0.0473 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0895 - lr: 1.0000e-05 - 202ms/epoch - 5ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0034 - mse: 0.0034 - mae: 0.0456 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0893 - lr: 1.0000e-05 - 220ms/epoch - 5ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0037 - mse: 0.0037 - mae: 0.0477 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0895 - lr: 1.0000e-05 - 209ms/epoch - 5ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0038 - mse: 0.0038 - mae: 0.0478 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0896 - lr: 1.0000e-05 - 226ms/epoch - 5ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0037 - mse: 0.0037 - mae: 0.0470 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0895 - lr: 1.0000e-05 - 262ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0039 - mse: 0.0039 - mae: 0.0482 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0895 - lr: 1.0000e-05 - 197ms/epoch - 5ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0037 - mse: 0.0037 - mae: 0.0475 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0894 - lr: 1.0000e-05 - 245ms/epoch - 6ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0035 - mse: 0.0035 - mae: 0.0467 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0895 - lr: 1.0000e-05 - 219ms/epoch - 5ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0037 - mse: 0.0037 - mae: 0.0477 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0893 - lr: 1.0000e-05 - 212ms/epoch - 5ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0038 - mse: 0.0038 - mae: 0.0474 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0892 - lr: 1.0000e-05 - 185ms/epoch - 4ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0034 - mse: 0.0034 - mae: 0.0451 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0893 - lr: 1.0000e-05 - 196ms/epoch - 5ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0039 - mse: 0.0039 - mae: 0.0483 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0893 - lr: 1.0000e-05 - 196ms/epoch - 5ms/step
Epoch 56/500
Epoch 00056: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0037 - mse: 0.0037 - mae: 0.0474 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0894 - lr: 1.0000e-05 - 222ms/epoch - 5ms/step
Epoch 57/500
Epoch 00057: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0035 - mse: 0.0035 - mae: 0.0473 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0894 - lr: 1.0000e-05 - 203ms/epoch - 5ms/step
Epoch 58/500
Epoch 00058: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0035 - mse: 0.0035 - mae: 0.0465 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0895 - lr: 1.0000e-05 - 219ms/epoch - 5ms/step
Epoch 59/500
Epoch 00059: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0040 - mse: 0.0040 - mae: 0.0482 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0898 - lr: 1.0000e-05 - 229ms/epoch - 5ms/step
Epoch 60/500
Epoch 00060: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0036 - mse: 0.0036 - mae: 0.0467 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0896 - lr: 1.0000e-05 - 214ms/epoch - 5ms/step
Epoch 61/500
Epoch 00061: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0033 - mse: 0.0033 - mae: 0.0444 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0896 - lr: 1.0000e-05 - 219ms/epoch - 5ms/step
Epoch 62/500
Epoch 00062: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0035 - mse: 0.0035 - mae: 0.0465 - val_loss: 0.0111 - val_mse: 0.0111 - val_mae: 0.0893 - lr: 1.0000e-05 - 219ms/epoch - 5ms/step
Epoch 63/500
Epoch 00063: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0036 - mse: 0.0036 - mae: 0.0463 - val_loss: 0.0112 - val_mse: 0.0112 - val_mae: 0.0895 - lr: 1.0000e-05 - 235ms/epoch - 5ms/step
Epoch 64/500
Epoch 00064: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0034 - mse: 0.0034 - mae: 0.0448 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0898 - lr: 1.0000e-05 - 191ms/epoch - 4ms/step
Epoch 65/500
Epoch 00065: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0033 - mse: 0.0033 - mae: 0.0448 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0900 - lr: 1.0000e-05 - 196ms/epoch - 5ms/step
Epoch 66/500
Epoch 00066: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0034 - mse: 0.0034 - mae: 0.0448 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0900 - lr: 1.0000e-05 - 220ms/epoch - 5ms/step
Epoch 67/500
Epoch 00067: val_loss did not improve from 0.00834
43/43 - 0s - loss: 0.0033 - mse: 0.0033 - mae: 0.0442 - val_loss: 0.0113 - val_mse: 0.0113 - val_mae: 0.0901 - lr: 1.0000e-05 - 184ms/epoch - 4ms/step
Epoch 00067: early stopping
SMA
Prediction vs Close: 50.0% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 133.6684179910327
RMSE: 11.561505870388714
MAPE: 10.289389775089397
EMA
Prediction vs Close: 50.0% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 34.52931938659473
RMSE: 5.876165364129462
MAPE: 4.852473639818076
WMA
Prediction vs Close: 51.49% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 36.74468487644727
RMSE: 6.061739426637149
MAPE: 4.85767480758183
DEMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 87.45802496937279
RMSE: 9.351899538028238
MAPE: 8.239361009856534
KAMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 55.655998887145394
RMSE: 7.460294825752223
MAPE: 6.325008398714769
MIDPOINT
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 43.66% Accuracy
MSE: 28.58496814832563
RMSE: 5.346491199686541
MAPE: 4.412567377496203
T3
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 44.78% Accuracy
MSE: 127.72960460138046
RMSE: 11.301752280128088
MAPE: 9.16046603172084
TEMA
TEMA([input_arrays], [timeperiod=30])
Triple Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
9
Working on TEMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16762.799, Time=4.57 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14158.507, Time=2.71 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-16445.598, Time=8.48 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-16144.282, Time=10.93 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-15275.101, Time=9.23 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-15897.090, Time=12.93 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-16446.973, Time=9.53 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=-16567.628, Time=3.34 sec
/usr/local/lib/python3.7/dist-packages/statsmodels/tsa/statespace/sarimax.py:1906: RuntimeWarning: divide by zero encountered in reciprocal return np.roots(self.polynomial_reduced_ma)**-1
ARIMA(2,3,2)(0,0,0)[0] : AIC=-16523.926, Time=3.65 sec
ARIMA(1,3,1)(0,0,0)[0] intercept : AIC=-16696.008, Time=3.22 sec
Best model: ARIMA(1,3,1)(0,0,0)[0]
Total fit time: 68.599 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(1, 3, 1) Log Likelihood 8408.400
Date: Sun, 12 Dec 2021 AIC -16762.799
Time: 17:50:11 BIC -16636.147
Sample: 0 HQIC -16714.159
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -5.289e-07 0.001 -0.000 1.000 -0.002 0.002
x2 -5.288e-07 0.001 -0.001 0.999 -0.002 0.002
x3 -5.306e-07 0.001 -0.000 1.000 -0.002 0.002
x4 1.0000 0.000 2045.695 0.000 0.999 1.001
x5 -5.041e-07 0.000 -0.001 0.999 -0.001 0.001
x6 -9.879e-07 4.33e-05 -0.023 0.982 -8.58e-05 8.38e-05
x7 -5.185e-07 0.001 -0.001 0.999 -0.001 0.001
x8 0.0001 0.000 0.643 0.520 -0.000 0.001
x9 9.794e-08 0.001 0.000 1.000 -0.001 0.001
x10 0.0001 0.000 0.313 0.754 -0.001 0.001
x11 -0.0004 0.000 -2.284 0.022 -0.001 -6.06e-05
x12 0.0005 0.000 2.453 0.014 0.000 0.001
x13 -5.277e-07 0.000 -0.002 0.999 -0.001 0.001
x14 -1.566e-06 0.000 -0.005 0.996 -0.001 0.001
x15 -5.136e-07 9.86e-05 -0.005 0.996 -0.000 0.000
x16 -7.66e-07 0.000 -0.002 0.999 -0.001 0.001
x17 -5.146e-07 0.000 -0.003 0.998 -0.000 0.000
x18 -1.701e-07 0.001 -0.000 1.000 -0.001 0.001
x19 -5.77e-07 8.54e-05 -0.007 0.995 -0.000 0.000
x20 5.026e-07 0.001 0.001 0.999 -0.001 0.001
x21 -2.058e-06 0.000 -0.010 0.992 -0.000 0.000
x22 -1.098e-06 0.001 -0.001 0.999 -0.003 0.003
x23 -1.472e-06 0.001 -0.003 0.998 -0.001 0.001
x24 -8.255e-07 0.001 -0.001 0.999 -0.002 0.002
ar.L1 -0.2866 3.63e-05 -7897.273 0.000 -0.287 -0.287
ma.L1 -0.9124 1.46e-06 -6.25e+05 0.000 -0.912 -0.912
sigma2 9.98e-11 7.23e-11 1.380 0.168 -4.2e-11 2.42e-10
===================================================================================
Ljung-Box (L1) (Q): 83.51 Jarque-Bera (JB): 4742889.91
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: -5.71
Prob(H) (two-sided): 0.00 Kurtosis: 378.86
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 4.2e+22. Standard errors may be unstable.
ARIMA order: (1, 3, 1)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.07750, saving model to LSTM7.h5
90/90 - 3s - loss: 0.1902 - mse: 0.1902 - mae: 0.2954 - val_loss: 0.0775 - val_mse: 0.0775 - val_mae: 0.2486 - lr: 0.0010 - 3s/epoch - 34ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.07750
90/90 - 0s - loss: 0.0241 - mse: 0.0241 - mae: 0.1217 - val_loss: 0.1034 - val_mse: 0.1034 - val_mae: 0.2967 - lr: 0.0010 - 382ms/epoch - 4ms/step
Epoch 3/500
Epoch 00003: val_loss improved from 0.07750 to 0.06287, saving model to LSTM7.h5
90/90 - 0s - loss: 0.0182 - mse: 0.0182 - mae: 0.1045 - val_loss: 0.0629 - val_mse: 0.0629 - val_mae: 0.2273 - lr: 0.0010 - 372ms/epoch - 4ms/step
Epoch 4/500
Epoch 00004: val_loss improved from 0.06287 to 0.06190, saving model to LSTM7.h5
90/90 - 0s - loss: 0.0140 - mse: 0.0140 - mae: 0.0924 - val_loss: 0.0619 - val_mse: 0.0619 - val_mae: 0.2270 - lr: 0.0010 - 383ms/epoch - 4ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0103 - mse: 0.0103 - mae: 0.0798 - val_loss: 0.0712 - val_mse: 0.0712 - val_mae: 0.2454 - lr: 0.0010 - 353ms/epoch - 4ms/step
Epoch 6/500
Epoch 00006: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0084 - mse: 0.0084 - mae: 0.0726 - val_loss: 0.0844 - val_mse: 0.0844 - val_mae: 0.2703 - lr: 0.0010 - 403ms/epoch - 4ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0079 - mse: 0.0079 - mae: 0.0701 - val_loss: 0.0739 - val_mse: 0.0739 - val_mae: 0.2512 - lr: 0.0010 - 374ms/epoch - 4ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0079 - mse: 0.0079 - mae: 0.0703 - val_loss: 0.1104 - val_mse: 0.1104 - val_mae: 0.3121 - lr: 0.0010 - 440ms/epoch - 5ms/step
Epoch 9/500
Epoch 00009: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00009: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0080 - mse: 0.0080 - mae: 0.0709 - val_loss: 0.1088 - val_mse: 0.1088 - val_mae: 0.3094 - lr: 0.0010 - 372ms/epoch - 4ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0174 - mse: 0.0174 - mae: 0.1065 - val_loss: 0.0653 - val_mse: 0.0653 - val_mae: 0.2330 - lr: 1.0000e-04 - 353ms/epoch - 4ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0081 - mse: 0.0081 - mae: 0.0716 - val_loss: 0.0680 - val_mse: 0.0680 - val_mae: 0.2385 - lr: 1.0000e-04 - 384ms/epoch - 4ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0068 - mse: 0.0068 - mae: 0.0652 - val_loss: 0.0700 - val_mse: 0.0700 - val_mae: 0.2423 - lr: 1.0000e-04 - 399ms/epoch - 4ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0069 - mse: 0.0069 - mae: 0.0652 - val_loss: 0.0715 - val_mse: 0.0715 - val_mae: 0.2450 - lr: 1.0000e-04 - 404ms/epoch - 4ms/step
Epoch 14/500
Epoch 00014: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00014: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0061 - mse: 0.0061 - mae: 0.0617 - val_loss: 0.0725 - val_mse: 0.0725 - val_mae: 0.2467 - lr: 1.0000e-04 - 382ms/epoch - 4ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0054 - mse: 0.0054 - mae: 0.0583 - val_loss: 0.0730 - val_mse: 0.0730 - val_mae: 0.2478 - lr: 1.0000e-05 - 349ms/epoch - 4ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0057 - mse: 0.0057 - mae: 0.0591 - val_loss: 0.0734 - val_mse: 0.0734 - val_mae: 0.2485 - lr: 1.0000e-05 - 370ms/epoch - 4ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0053 - mse: 0.0053 - mae: 0.0563 - val_loss: 0.0739 - val_mse: 0.0739 - val_mae: 0.2494 - lr: 1.0000e-05 - 356ms/epoch - 4ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0049 - mse: 0.0049 - mae: 0.0551 - val_loss: 0.0744 - val_mse: 0.0744 - val_mae: 0.2503 - lr: 1.0000e-05 - 361ms/epoch - 4ms/step
Epoch 19/500
Epoch 00019: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00019: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0054 - mse: 0.0054 - mae: 0.0576 - val_loss: 0.0743 - val_mse: 0.0743 - val_mae: 0.2503 - lr: 1.0000e-05 - 479ms/epoch - 5ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0566 - val_loss: 0.0750 - val_mse: 0.0750 - val_mae: 0.2515 - lr: 1.0000e-05 - 374ms/epoch - 4ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0053 - mse: 0.0053 - mae: 0.0578 - val_loss: 0.0753 - val_mse: 0.0753 - val_mae: 0.2522 - lr: 1.0000e-05 - 387ms/epoch - 4ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0048 - mse: 0.0048 - mae: 0.0561 - val_loss: 0.0756 - val_mse: 0.0756 - val_mae: 0.2527 - lr: 1.0000e-05 - 400ms/epoch - 4ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0053 - mse: 0.0053 - mae: 0.0569 - val_loss: 0.0757 - val_mse: 0.0757 - val_mae: 0.2528 - lr: 1.0000e-05 - 379ms/epoch - 4ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0049 - mse: 0.0049 - mae: 0.0551 - val_loss: 0.0758 - val_mse: 0.0758 - val_mae: 0.2530 - lr: 1.0000e-05 - 396ms/epoch - 4ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0559 - val_loss: 0.0756 - val_mse: 0.0756 - val_mae: 0.2527 - lr: 1.0000e-05 - 394ms/epoch - 4ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0552 - val_loss: 0.0758 - val_mse: 0.0758 - val_mae: 0.2530 - lr: 1.0000e-05 - 484ms/epoch - 5ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0052 - mse: 0.0052 - mae: 0.0566 - val_loss: 0.0759 - val_mse: 0.0759 - val_mae: 0.2532 - lr: 1.0000e-05 - 381ms/epoch - 4ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0524 - val_loss: 0.0765 - val_mse: 0.0765 - val_mae: 0.2543 - lr: 1.0000e-05 - 353ms/epoch - 4ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0052 - mse: 0.0052 - mae: 0.0570 - val_loss: 0.0767 - val_mse: 0.0767 - val_mae: 0.2547 - lr: 1.0000e-05 - 370ms/epoch - 4ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0054 - mse: 0.0054 - mae: 0.0574 - val_loss: 0.0764 - val_mse: 0.0764 - val_mae: 0.2541 - lr: 1.0000e-05 - 364ms/epoch - 4ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0049 - mse: 0.0049 - mae: 0.0556 - val_loss: 0.0771 - val_mse: 0.0771 - val_mae: 0.2553 - lr: 1.0000e-05 - 347ms/epoch - 4ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0549 - val_loss: 0.0778 - val_mse: 0.0778 - val_mae: 0.2566 - lr: 1.0000e-05 - 358ms/epoch - 4ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0051 - mse: 0.0051 - mae: 0.0562 - val_loss: 0.0775 - val_mse: 0.0775 - val_mae: 0.2561 - lr: 1.0000e-05 - 365ms/epoch - 4ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0043 - mse: 0.0043 - mae: 0.0520 - val_loss: 0.0776 - val_mse: 0.0776 - val_mae: 0.2563 - lr: 1.0000e-05 - 338ms/epoch - 4ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0545 - val_loss: 0.0781 - val_mse: 0.0781 - val_mae: 0.2571 - lr: 1.0000e-05 - 364ms/epoch - 4ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0045 - mse: 0.0045 - mae: 0.0529 - val_loss: 0.0783 - val_mse: 0.0783 - val_mae: 0.2575 - lr: 1.0000e-05 - 373ms/epoch - 4ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0538 - val_loss: 0.0783 - val_mse: 0.0783 - val_mae: 0.2575 - lr: 1.0000e-05 - 362ms/epoch - 4ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0527 - val_loss: 0.0789 - val_mse: 0.0789 - val_mae: 0.2586 - lr: 1.0000e-05 - 352ms/epoch - 4ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0043 - mse: 0.0043 - mae: 0.0522 - val_loss: 0.0798 - val_mse: 0.0798 - val_mae: 0.2603 - lr: 1.0000e-05 - 345ms/epoch - 4ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0039 - mse: 0.0039 - mae: 0.0494 - val_loss: 0.0796 - val_mse: 0.0796 - val_mae: 0.2598 - lr: 1.0000e-05 - 387ms/epoch - 4ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0512 - val_loss: 0.0799 - val_mse: 0.0799 - val_mae: 0.2604 - lr: 1.0000e-05 - 394ms/epoch - 4ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0043 - mse: 0.0043 - mae: 0.0515 - val_loss: 0.0794 - val_mse: 0.0794 - val_mae: 0.2595 - lr: 1.0000e-05 - 348ms/epoch - 4ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0535 - val_loss: 0.0796 - val_mse: 0.0796 - val_mae: 0.2598 - lr: 1.0000e-05 - 355ms/epoch - 4ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0045 - mse: 0.0045 - mae: 0.0520 - val_loss: 0.0800 - val_mse: 0.0800 - val_mae: 0.2605 - lr: 1.0000e-05 - 350ms/epoch - 4ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0048 - mse: 0.0048 - mae: 0.0534 - val_loss: 0.0805 - val_mse: 0.0805 - val_mae: 0.2615 - lr: 1.0000e-05 - 357ms/epoch - 4ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0047 - mse: 0.0047 - mae: 0.0543 - val_loss: 0.0792 - val_mse: 0.0792 - val_mae: 0.2590 - lr: 1.0000e-05 - 363ms/epoch - 4ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0515 - val_loss: 0.0785 - val_mse: 0.0785 - val_mae: 0.2578 - lr: 1.0000e-05 - 354ms/epoch - 4ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0522 - val_loss: 0.0788 - val_mse: 0.0788 - val_mae: 0.2583 - lr: 1.0000e-05 - 371ms/epoch - 4ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0041 - mse: 0.0041 - mae: 0.0506 - val_loss: 0.0802 - val_mse: 0.0802 - val_mae: 0.2607 - lr: 1.0000e-05 - 351ms/epoch - 4ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0042 - mse: 0.0042 - mae: 0.0512 - val_loss: 0.0797 - val_mse: 0.0797 - val_mae: 0.2598 - lr: 1.0000e-05 - 355ms/epoch - 4ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0040 - mse: 0.0040 - mae: 0.0484 - val_loss: 0.0796 - val_mse: 0.0796 - val_mae: 0.2597 - lr: 1.0000e-05 - 348ms/epoch - 4ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0045 - mse: 0.0045 - mae: 0.0525 - val_loss: 0.0804 - val_mse: 0.0804 - val_mae: 0.2611 - lr: 1.0000e-05 - 364ms/epoch - 4ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0044 - mse: 0.0044 - mae: 0.0523 - val_loss: 0.0783 - val_mse: 0.0783 - val_mae: 0.2573 - lr: 1.0000e-05 - 434ms/epoch - 5ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.06190
90/90 - 0s - loss: 0.0046 - mse: 0.0046 - mae: 0.0530 - val_loss: 0.0777 - val_mse: 0.0777 - val_mae: 0.2562 - lr: 1.0000e-05 - 389ms/epoch - 4ms/step
Epoch 00054: early stopping
SMA Prediction vs Close: 50.0% Accuracy Prediction vs Prediction: 48.51% Accuracy MSE: 133.6684179910327 RMSE: 11.561505870388714 MAPE: 10.289389775089397 EMA Prediction vs Close: 50.0% Accuracy Prediction vs Prediction: 45.9% Accuracy MSE: 34.52931938659473 RMSE: 5.876165364129462 MAPE: 4.852473639818076 WMA Prediction vs Close: 51.49% Accuracy Prediction vs Prediction: 47.01% Accuracy MSE: 36.74468487644727 RMSE: 6.061739426637149 MAPE: 4.85767480758183 DEMA Prediction vs Close: 54.1% Accuracy Prediction vs Prediction: 45.52% Accuracy MSE: 87.45802496937279 RMSE: 9.351899538028238 MAPE: 8.239361009856534 KAMA Prediction vs Close: 52.99% Accuracy Prediction vs Prediction: 45.52% Accuracy MSE: 55.655998887145394 RMSE: 7.460294825752223 MAPE: 6.325008398714769 MIDPOINT Prediction vs Close: 52.99% Accuracy Prediction vs Prediction: 43.66% Accuracy MSE: 28.58496814832563 RMSE: 5.346491199686541 MAPE: 4.412567377496203 T3 Prediction vs Close: 52.99% Accuracy Prediction vs Prediction: 44.78% Accuracy MSE: 127.72960460138046 RMSE: 11.301752280128088 MAPE: 9.16046603172084 TEMA Prediction vs Close: 49.63% Accuracy Prediction vs Prediction: 49.25% Accuracy MSE: 34.19976389234231 RMSE: 5.8480564200717415 MAPE: 5.017065986818192 Runtime: mins: 53.440131934050015
Architecture Used¶
In [134]:
from google.colab import files
import cv2
uploaded = files.upload()
Saving Experiment7.png to Experiment7 (1).png
In [135]:
img = cv2.imread('Experiment5.png')
plt.figure(figsize=(20,10))
plt.axis("off")
plt.title('LSTM Architecture '+imgfile,fontsize=18)
plt.imshow(img)
Out[135]:
<matplotlib.image.AxesImage at 0x7f75dc198a90>
Model Plots¶
In [110]:
with open('simulation7_data.json') as json_file:
simulation7 = json.load(json_file)
fileimg = 'Experiment7'
In [111]:
SIM = list(simulation7.keys())[i]
plot_train(simulation7,SIM)
plot_test(simulation7,SIM)
----- Train RMSE for TEMA ----- 7.535421816948624 ----- Train_MSE_LSTM for TEMA ----- 56.7825819593453 ----- Train MAE LSTM for TEMA ----- 5.2132834616768555
----- Test RMSE for TEMA----- 5.8480564200717415 ----- Test_MSE_LSTM for TEMA----- 34.19976389234231 ----- Test_MAE_LSTM for TEMA----- 5.017065986818192
Arima w Exogenous Variable Multistep MutiVariate LSTM Hybrid Model Experiment 8¶
In [137]:
def get_arima_exog(dataframe,original_data, train_len, test_len):
# prepare train and test data for exogenous vr
X_value = pd.DataFrame(low_vol.iloc[:, :])
y_value = pd.DataFrame(low_vol.iloc[:, 3])
X_scaler = MinMaxScaler(feature_range=(-1, 1))
y_scaler = MinMaxScaler(feature_range=(-1, 1))
X_scaler.fit(X_value)
y_scaler.fit(y_value)
X_scale_dataset = X_scaler.fit_transform(X_value)
y_scale_dataset = y_scaler.fit_transform(y_value)
# Get data and check shape
# X, y, yc = get_X_y(X_scale_dataset, y_scale_dataset)#X will be of shape 224 X 3 X 21 (each 3 X 21 array will be 3 days' worth of data). yc will have the corresponding closing price value
# pdb.set_trace()
X_train, X_test, = split_train_test(X_scale_dataset)
y_train, y_test, = split_train_test(y_scale_dataset)
yc_train,yc_test = split_train_test(low_vol_data)
yc = yc_test.values.tolist()
y_train_list = y_train.flatten().tolist()
y_test_list = y_test.flatten().tolist()
# yc_train, yc_test, = split_train_test(original_data)
index_train, index_test, = predict_index(dataset_final, X_train, n_steps_in, n_steps_out)
# Initialize model
model = auto_arima(y_train_list,exogenous = X_train,trace=True, error_action='ignore', start_p=1,start_q=1,max_p=3,max_q=3,d=3,
suppress_warnings=True,stepwise=True,seasonal=True)
# Determine model parameters
print(model.summary())
model.fit(y_train_list,maxiter=200)
order = model.get_params()['order']
print('ARIMA order:', order, '\n')
# Genereate predictions
prediction = []
for i in range(len(y_test_list)):
model = pmdarima.ARIMA(order=order)
model.fit(y_train_list)
# print('working on', i+1, 'of', len(y_test), '-- ' + str(int(100 * (i + 1) / len(y_test))) + '% complete')
prediction.append(model.predict()[0])
y_train_list.append(y_test_list[i])
predictionte = y_scaler.inverse_transform(np.array(prediction).reshape(-1,1))
y_test_ = y_scaler.inverse_transform(np.array(y_test_list).reshape(-1,1))
# Generate error data
mse = mean_squared_error(yc_test, predictionte)
rmse = mse ** 0.5
mae = mean_absolute_error(y_test_ , predictionte )
return yc,predictionte.flatten().tolist(), mse, rmse, mae
In [138]:
def get_lstm(data,original_data, train_len, test_len,img_file,ma ,lstm_len=3):
# prepare train and test data
X_value = pd.DataFrame(data.iloc[:, :])
y_value = pd.DataFrame(data.iloc[:, 3])
X_scaler = MinMaxScaler(feature_range=(-1, 1))
y_scaler = MinMaxScaler(feature_range=(-1, 1))
X_scaler.fit(X_value)
y_scaler.fit(y_value)
# Get data and check shape
X, y, yc = get_X_y(X_scale_dataset, y_scale_dataset)#X will be of shape 224 X 3 X 21 (each 3 X 21 array will be 3 days' worth of data). yc will have the corresponding closing price value
# pdb.set_trace()
X_train, X_test, = split_train_test(X)
y_train, y_test, = split_train_test(y)
# yc_train, yc_test, = split_train_test(original_data)
index_train, index_test, = predict_index(dataset_final, X_train, n_steps_in, n_steps_out)
det =20
input_dim = X_train.shape[1]#3
feature_size = X_train.shape[2]#24
output_dim = y_train.shape[1]#1
# Option 1
# Set up & fit LSTM RNN
# model = Sequential()
# model.add(LSTM(256, activation='relu', kernel_initializer='he_normal', input_shape=(input_dim, feature_size)))
# model.add(Dense(units=64,activation='relu'))
# model.add(Dropout(0.5))
# model.add(Dense(units=output_dim))
# model.compile(optimizer=Adam(learning_rate = 0.001), loss='mse')
# ## Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM1.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma_' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# # # option 2
# model = Sequential()
# model.add(Bidirectional(LSTM(units= 128), input_shape=(input_dim, feature_size)))
# model.add(Dense(64))
# model.add(Dense(units=output_dim))
# model.compile(optimizer=Adam(lr = 0.001), loss='mean_squared_error', metrics=['accuracy'])
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM7.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma+' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# Option 3
# define custom activation
#
# class Double_Tanh(Activation):
# def __init__(self, activation, **kwargs):
# super(Double_Tanh, self).__init__(activation, **kwargs)
# self.__name__ = 'double_tanh'
# def double_tanh(x):
# return (K.tanh(x) * 2)
# get_custom_objects().update({'double_tanh':Double_Tanh(double_tanh)})
# # Model Generation
# model = Sequential()
# #check https://machinelearningmastery.com/use-weight-regularization-lstm-networks-time-series-forecasting/
# model.add(LSTM(25, input_shape=(input_dim, feature_size), dropout=0.2, kernel_regularizer=l1_l2(0.00,0.00), bias_regularizer=l1_l2(0.00,0.00)))
# model.add(Dense(1))
# model.add(Activation(double_tanh))
# model.compile(loss='mean_squared_error', optimizer='adam', metrics=['mse', 'mae'])
# # Common code
# callbacks = [
# EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
# ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
# ModelCheckpoint('LSTM7.h5', verbose=1, save_best_only=True, save_weights_only=True)]
# fname1 = img_file+'.png'
# tensorflow.keras.utils.plot_model(
# model, to_file=fname1, show_shapes=True, show_dtype=False,
# show_layer_names=True, expand_nested=False, dpi=96,
# layer_range=None, show_layer_activations=False
# )
# history = model.fit(X_train, y_train, epochs=500, batch_size=1, verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# # plot loss
# fname2 = img_file+'-'+ma
# plt.title(img_file+'-'+ma+' Loss')
# plt.xlabel("Epochs")
# plt.ylabel("Loss")
# pyplot.plot(history.history['loss'], label='train')
# pyplot.plot(history.history['val_loss'], label='validation')
# pyplot.legend()
# pyplot.savefig(fname2+'.png',dpi='figure')
# pyplot.show()
# #Option 4
# # Set up & fit LSTM RNN
model = Sequential()
model.add(LSTM(units=lstm_len, return_sequences=True, input_shape=(input_dim, feature_size)))
model.add(LSTM(units=int(lstm_len/2)))
model.add(Dense(1, activation='sigmoid'))
model.compile(loss='mean_squared_error', optimizer='adam')
# Common code
callbacks = [
EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=50),
ReduceLROnPlateau(factor=0.1, patience=5, min_lr=0.00001, verbose=1),
ModelCheckpoint('LSTM8.h5', verbose=1, save_best_only=True, save_weights_only=True)]
fname1 = img_file+'.png'
tensorflow.keras.utils.plot_model(
model, to_file=fname1, show_shapes=True, show_dtype=False,
show_layer_names=True, expand_nested=False, dpi=96,
layer_range=None, show_layer_activations=False
)
history = model.fit(X_train, y_train, epochs=500, batch_size=int( optimized_period[ma]), verbose=2, callbacks=callbacks, validation_data=(X_test, y_test),shuffle=False)
# plot loss
fname2 = img_file+'-'+ma
plt.title(img_file+'-'+ma+' Loss')
plt.xlabel("Epochs")
plt.ylabel("Loss")
pyplot.plot(history.history['loss'], label='train')
pyplot.plot(history.history['val_loss'], label='validation')
pyplot.legend()
pyplot.savefig(fname2+'.png',dpi='figure')
pyplot.show()
# Generate predictions
predictiontr = model.predict(X_train, verbose=0)
predictiontr = y_scaler.inverse_transform(predictiontr).tolist()
outputtr = []
for i in range(len(predictiontr)):
outputtr.extend(predictiontr[i])
predictiontr = outputtr
# Generate error data
## replace with yc , xtest generated by new multistep method
mse_tr = mean_squared_error(y_train, predictiontr)
rmse_tr = mse_tr ** 0.5
# mape = mean_absolute_percentage_error(X_test, pd.Series(predictiontr))
mae_tr = mean_absolute_error(y_train, pd.Series(predictiontr))
# Original_tr = pd.Series(yc_train)
Original_tr = y_scaler.inverse_transform(y_train).flatten().tolist()
predictionte = model.predict(X_test, verbose=0)
predictionte = (y_scaler.inverse_transform(predictionte)-det).tolist()
outputte = []
for i in range(len(predictionte)):
outputte.extend(predictionte[i])
predictionte = outputte
# Generate error data
mse_te = mean_squared_error(y_test, predictionte)
rmse_te = mse_te ** 0.5
# mape = mean_absolute_percentage_error(X_test, pd.Series(predictionte))
mae_te = mean_absolute_error(y_test, pd.Series(predictionte))
# Original_te = pd.Series(yc_test)
Original_te = y_scaler.inverse_transform(y_test).flatten().tolist()
return Original_tr, predictiontr, mse_tr, rmse_tr,mae_tr,Original_te,predictionte, mse_te,rmse_te,mae_te
In [139]:
if __name__ == '__main__':
start_time = timeit.default_timer()
simulation8 = {}
imgfile = 'Experiment8'
for ma in optimized_period:
print(ma)
print(functions[ma])
print ( int( optimized_period[ma]))
# if ma == 'SMA':
low_vol = df.apply(lambda c: functions[ma](c, timeperiod = int( optimized_period[ma])))
low_vol = low_vol.fillna(0)
low_vol_data = df['close']
high_vol = pd.DataFrame()
df2 = df.copy()
for i in df2.columns:
if i in low_vol.columns:
high_vol[i] = df2[i].subtract(low_vol[i], fill_value=0)
high_vol_data = df['close']
## *****************************************************
# Generate ARIMA and LSTM predictions
print('\nWorking on ' + ma + ' predictions')
try:
print('parameters used : ', train_len, test_len)
low_vol_Original, low_vol_prediction, low_vol_mse, low_vol_rmse,low_vol_mae = get_arima_exog(low_vol,low_vol_data, train_len, test_len)
except:
print('ARIMA error, skipping to next MA type')
continue
Original_tr, predictiontr, mse_tr, rmse_tr,mae_tr,high_vol_Original, high_vol_prediction, high_vol_mse, high_vol_rmse,high_vol_mae, = get_lstm(high_vol,high_vol_data, train_len, test_len,imgfile,ma)
final_prediction_tr = df['close'].head(train_len).values + pd.Series(predictiontr) # ignoring first 3 steps
mse_ftr = mean_squared_error(df['close'].head(train_len).values,final_prediction_tr.values)
rmse_ftr = mse_ftr ** 0.5
mape_ftr = mean_absolute_percentage_error(df['close'].head(train_len).reset_index(drop=True), final_prediction_tr)
mae_ftr = mean_absolute_error(df['close'].head(train_len).reset_index(drop=True), final_prediction_tr)
final_prediction = pd.Series(low_vol_prediction[3:]) + pd.Series(high_vol_prediction)
mse = mean_squared_error(df['close'].tail(test_len).values,final_prediction.values)
rmse = mse ** 0.5
mape = mean_absolute_percentage_error(df['close'].tail(test_len).reset_index(drop=True), final_prediction)
mae = mean_absolute_error(df['close'].tail(test_len).reset_index(drop=True), final_prediction)
# Generate prediction accuracy
actual = df['close'].tail(test_len).values
result_1 = []
result_2 = []
for i in range(1, len(final_prediction)):
# Compare prediction to previous close price
if final_prediction[i] > actual[i-1] and actual[i] > actual[i-1]:
result_1.append(1)
elif final_prediction[i] < actual[i-1] and actual[i] < actual[i-1]:
result_1.append(1)
else:
result_1.append(0)
# Compare prediction to previous prediction
if final_prediction[i] > final_prediction[i-1] and actual[i] > actual[i-1]:
result_2.append(1)
elif final_prediction[i] < final_prediction[i-1] and actual[i] < actual[i-1]:
result_2.append(1)
else:
result_2.append(0)
accuracy_1 = np.mean(result_1)
accuracy_2 = np.mean(result_2)
simulation8[ma] = {'low_vol': {'original':list(low_vol_Original), 'prediction': list(low_vol_prediction) , 'mse': low_vol_mse,
'rmse': low_vol_rmse, 'mae' : low_vol_mae},
'high_vol': {'original':list(high_vol_Original),'prediction': list(high_vol_prediction), 'mse': high_vol_mse,
'rmse': high_vol_rmse, 'mae' : high_vol_mae},
'final_tr': {'original':df['close'].head(train_len).tolist(),'prediction': final_prediction_tr.values.tolist(), 'mse': mse_ftr,
'rmse': rmse_ftr, 'mae' : mae_ftr},
'final': {'original': df['close'].tail(test_len).tolist(), 'prediction': final_prediction.values.tolist(), 'mse': mse,
'rmse': rmse, 'mae': mae },
'accuracy': {'prediction vs close': accuracy_1, 'prediction vs prediction': accuracy_2}}
# save simulation data here as checkpoint
with open('simulation8_data.json', 'w') as fp:
json.dump(simulation8, fp)
for ma in simulation8.keys():
print('\n' + ma)
print('Prediction vs Close:\t\t' + str(round(100*simulation8[ma]['accuracy']['prediction vs close'], 2))
+ '% Accuracy')
print('Prediction vs Prediction:\t' + str(round(100*simulation8[ma]['accuracy']['prediction vs prediction'], 2))
+ '% Accuracy')
print('MSE:\t', simulation8[ma]['final']['mse'],
'\nRMSE:\t', simulation8[ma]['final']['rmse'],
'\nMAPE:\t', simulation8[ma]['final']['mae'])#,
# '\nMAPE:\t', simulation[ma]['final']['mape'])
# else:
# break
elapsed = timeit.default_timer() - start_time
print('Runtime: mins:',elapsed/60)
SMA
SMA([input_arrays], [timeperiod=30])
Simple Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
17
Working on SMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.786, Time=3.28 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.592, Time=4.67 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-15578.394, Time=8.51 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.592, Time=7.38 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16966.361, Time=9.48 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-16121.635, Time=9.85 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17214.069, Time=12.68 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-14562.592, Time=9.21 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-14572.319, Time=9.59 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-14403.474, Time=41.33 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 115.999 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8634.035
Date: Sun, 12 Dec 2021 AIC -17214.069
Time: 18:02:30 BIC -17087.416
Sample: 0 HQIC -17165.429
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -4.257e-09 9.55e-06 -0.000 1.000 -1.87e-05 1.87e-05
x2 -4.256e-09 9.56e-06 -0.000 1.000 -1.87e-05 1.87e-05
x3 -4.313e-09 9.62e-06 -0.000 1.000 -1.89e-05 1.88e-05
x4 1.0000 9.61e-06 1.04e+05 0.000 1.000 1.000
x5 -3.891e-09 9.14e-06 -0.000 1.000 -1.79e-05 1.79e-05
x6 -1.122e-08 1.03e-05 -0.001 0.999 -2.03e-05 2.03e-05
x7 -4.223e-09 9.54e-06 -0.000 1.000 -1.87e-05 1.87e-05
x8 -4.234e-09 9.55e-06 -0.000 1.000 -1.87e-05 1.87e-05
x9 -1.626e-10 6.54e-07 -0.000 1.000 -1.28e-06 1.28e-06
x10 -6.831e-10 2.91e-06 -0.000 1.000 -5.7e-06 5.7e-06
x11 -4.115e-09 9.41e-06 -0.000 1.000 -1.84e-05 1.84e-05
x12 -4.303e-09 9.62e-06 -0.000 1.000 -1.89e-05 1.88e-05
x13 -4.288e-09 9.6e-06 -0.000 1.000 -1.88e-05 1.88e-05
x14 -3.749e-08 2.81e-05 -0.001 0.999 -5.51e-05 5.5e-05
x15 -5.032e-09 1.04e-05 -0.000 1.000 -2.04e-05 2.03e-05
x16 -3.685e-09 9e-06 -0.000 1.000 -1.76e-05 1.76e-05
x17 -3.286e-09 8.45e-06 -0.000 1.000 -1.66e-05 1.66e-05
x18 -1.22e-08 1.59e-05 -0.001 0.999 -3.11e-05 3.11e-05
x19 -5.685e-09 1.1e-05 -0.001 1.000 -2.16e-05 2.16e-05
x20 -1.42e-08 1.69e-05 -0.001 0.999 -3.32e-05 3.32e-05
x21 -5.194e-08 3.31e-05 -0.002 0.999 -6.49e-05 6.48e-05
x22 -2.548e-08 2.31e-05 -0.001 0.999 -4.53e-05 4.52e-05
x23 -3.534e-08 2.73e-05 -0.001 0.999 -5.35e-05 5.34e-05
x24 -1.566e-08 1.8e-05 -0.001 0.999 -3.53e-05 3.53e-05
ma.L1 -1.3899 4.98e-09 -2.79e+08 0.000 -1.390 -1.390
ma.L2 0.4032 4.98e-09 8.09e+07 0.000 0.403 0.403
sigma2 7.635e-11 6.92e-11 1.103 0.270 -5.93e-11 2.12e-10
===================================================================================
Ljung-Box (L1) (Q): 68.48 Jarque-Bera (JB): 5579791.06
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 10.12
Prob(H) (two-sided): 0.00 Kurtosis: 410.36
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.69e+24. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.04823, saving model to LSTM8.h5
48/48 - 4s - loss: 1.3964 - val_loss: 0.0482 - lr: 0.0010 - 4s/epoch - 77ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.04823
48/48 - 0s - loss: 1.3224 - val_loss: 0.0519 - lr: 0.0010 - 287ms/epoch - 6ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.04823
48/48 - 0s - loss: 1.2289 - val_loss: 0.0582 - lr: 0.0010 - 277ms/epoch - 6ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.04823
48/48 - 0s - loss: 1.1176 - val_loss: 0.0641 - lr: 0.0010 - 257ms/epoch - 5ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.04823
48/48 - 0s - loss: 1.0133 - val_loss: 0.0698 - lr: 0.0010 - 239ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00006: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.9452 - val_loss: 0.0754 - lr: 0.0010 - 273ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.9122 - val_loss: 0.0760 - lr: 1.0000e-04 - 286ms/epoch - 6ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.9077 - val_loss: 0.0765 - lr: 1.0000e-04 - 232ms/epoch - 5ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.9032 - val_loss: 0.0771 - lr: 1.0000e-04 - 228ms/epoch - 5ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8988 - val_loss: 0.0777 - lr: 1.0000e-04 - 251ms/epoch - 5ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00011: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8946 - val_loss: 0.0783 - lr: 1.0000e-04 - 277ms/epoch - 6ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8919 - val_loss: 0.0784 - lr: 1.0000e-05 - 239ms/epoch - 5ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8915 - val_loss: 0.0785 - lr: 1.0000e-05 - 245ms/epoch - 5ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8911 - val_loss: 0.0785 - lr: 1.0000e-05 - 244ms/epoch - 5ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8907 - val_loss: 0.0786 - lr: 1.0000e-05 - 259ms/epoch - 5ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00016: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8902 - val_loss: 0.0787 - lr: 1.0000e-05 - 234ms/epoch - 5ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8898 - val_loss: 0.0787 - lr: 1.0000e-05 - 235ms/epoch - 5ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8894 - val_loss: 0.0788 - lr: 1.0000e-05 - 263ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8889 - val_loss: 0.0789 - lr: 1.0000e-05 - 249ms/epoch - 5ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8885 - val_loss: 0.0790 - lr: 1.0000e-05 - 240ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8881 - val_loss: 0.0790 - lr: 1.0000e-05 - 236ms/epoch - 5ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8876 - val_loss: 0.0791 - lr: 1.0000e-05 - 251ms/epoch - 5ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8872 - val_loss: 0.0792 - lr: 1.0000e-05 - 265ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8868 - val_loss: 0.0793 - lr: 1.0000e-05 - 237ms/epoch - 5ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8863 - val_loss: 0.0793 - lr: 1.0000e-05 - 266ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8859 - val_loss: 0.0794 - lr: 1.0000e-05 - 311ms/epoch - 6ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8855 - val_loss: 0.0795 - lr: 1.0000e-05 - 269ms/epoch - 6ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8850 - val_loss: 0.0796 - lr: 1.0000e-05 - 234ms/epoch - 5ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8846 - val_loss: 0.0797 - lr: 1.0000e-05 - 238ms/epoch - 5ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8842 - val_loss: 0.0797 - lr: 1.0000e-05 - 261ms/epoch - 5ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8837 - val_loss: 0.0798 - lr: 1.0000e-05 - 275ms/epoch - 6ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8833 - val_loss: 0.0799 - lr: 1.0000e-05 - 252ms/epoch - 5ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8828 - val_loss: 0.0800 - lr: 1.0000e-05 - 266ms/epoch - 6ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8824 - val_loss: 0.0801 - lr: 1.0000e-05 - 269ms/epoch - 6ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8820 - val_loss: 0.0802 - lr: 1.0000e-05 - 250ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8815 - val_loss: 0.0803 - lr: 1.0000e-05 - 239ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8811 - val_loss: 0.0803 - lr: 1.0000e-05 - 270ms/epoch - 6ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8806 - val_loss: 0.0804 - lr: 1.0000e-05 - 264ms/epoch - 6ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8802 - val_loss: 0.0805 - lr: 1.0000e-05 - 254ms/epoch - 5ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8798 - val_loss: 0.0806 - lr: 1.0000e-05 - 277ms/epoch - 6ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8793 - val_loss: 0.0807 - lr: 1.0000e-05 - 243ms/epoch - 5ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8789 - val_loss: 0.0808 - lr: 1.0000e-05 - 271ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8785 - val_loss: 0.0809 - lr: 1.0000e-05 - 297ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8780 - val_loss: 0.0810 - lr: 1.0000e-05 - 290ms/epoch - 6ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8776 - val_loss: 0.0811 - lr: 1.0000e-05 - 251ms/epoch - 5ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8771 - val_loss: 0.0812 - lr: 1.0000e-05 - 264ms/epoch - 5ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8767 - val_loss: 0.0812 - lr: 1.0000e-05 - 263ms/epoch - 5ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8763 - val_loss: 0.0813 - lr: 1.0000e-05 - 272ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8758 - val_loss: 0.0814 - lr: 1.0000e-05 - 279ms/epoch - 6ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8754 - val_loss: 0.0815 - lr: 1.0000e-05 - 262ms/epoch - 5ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.04823
48/48 - 0s - loss: 0.8750 - val_loss: 0.0816 - lr: 1.0000e-05 - 239ms/epoch - 5ms/step
Epoch 00051: early stopping
SMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 23.82392201920814
RMSE: 4.880975519218278
MAPE: 3.848681389687446
EMA
EMA([input_arrays], [timeperiod=30])
Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
51
Working on EMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.778, Time=3.08 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.589, Time=4.43 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-14606.447, Time=6.05 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.589, Time=6.66 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-15343.613, Time=9.81 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-15047.583, Time=13.53 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-16858.964, Time=11.94 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=-17024.022, Time=6.03 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=-16998.618, Time=3.25 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=-17081.451, Time=6.60 sec
ARIMA(3,3,2)(0,0,0)[0] : AIC=inf, Time=16.83 sec
/usr/local/lib/python3.7/dist-packages/statsmodels/tsa/statespace/sarimax.py:1906: RuntimeWarning: divide by zero encountered in reciprocal return np.roots(self.polynomial_reduced_ma)**-1
ARIMA(2,3,2)(0,0,0)[0] : AIC=-16997.990, Time=3.51 sec
ARIMA(3,3,1)(0,0,0)[0] intercept : AIC=-16992.667, Time=4.34 sec
Best model: ARIMA(3,3,1)(0,0,0)[0]
Total fit time: 96.060 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 1) Log Likelihood 8569.726
Date: Sun, 12 Dec 2021 AIC -17081.451
Time: 18:08:04 BIC -16945.417
Sample: 0 HQIC -17029.208
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -2.316e-10 9.89e-05 -2.34e-06 1.000 -0.000 0.000
x2 -2.309e-10 9.88e-05 -2.34e-06 1.000 -0.000 0.000
x3 -2.325e-10 9.91e-05 -2.35e-06 1.000 -0.000 0.000
x4 1.0000 9.9e-05 1.01e+04 0.000 1.000 1.000
x5 -2.108e-10 9.43e-05 -2.24e-06 1.000 -0.000 0.000
x6 -7.997e-10 0.000 -4.63e-06 1.000 -0.000 0.000
x7 -2.295e-10 9.85e-05 -2.33e-06 1.000 -0.000 0.000
x8 -2.244e-10 9.74e-05 -2.3e-06 1.000 -0.000 0.000
x9 -1.166e-11 1.98e-05 -5.9e-07 1.000 -3.87e-05 3.87e-05
x10 -4.454e-11 4.19e-05 -1.06e-06 1.000 -8.22e-05 8.22e-05
x11 -2.219e-10 9.68e-05 -2.29e-06 1.000 -0.000 0.000
x12 -2.264e-10 9.8e-05 -2.31e-06 1.000 -0.000 0.000
x13 -2.315e-10 9.89e-05 -2.34e-06 1.000 -0.000 0.000
x14 -1.767e-09 0.000 -6.47e-06 1.000 -0.001 0.001
x15 -2.096e-10 9.38e-05 -2.23e-06 1.000 -0.000 0.000
x16 -5.257e-10 0.000 -3.5e-06 1.000 -0.000 0.000
x17 -2.143e-10 9.53e-05 -2.25e-06 1.000 -0.000 0.000
x18 -3.776e-11 3.61e-05 -1.05e-06 1.000 -7.08e-05 7.08e-05
x19 -2.52e-10 0.000 -2.41e-06 1.000 -0.000 0.000
x20 -2.417e-10 9.51e-05 -2.54e-06 1.000 -0.000 0.000
x21 -3.16e-09 0.000 -8.64e-06 1.000 -0.001 0.001
x22 -2.955e-09 0.000 -8.32e-06 1.000 -0.001 0.001
x23 -1.664e-09 0.000 -6.29e-06 1.000 -0.001 0.001
x24 -1.568e-09 0.000 -6.07e-06 1.000 -0.001 0.001
ar.L1 -0.4923 1.2e-09 -4.09e+08 0.000 -0.492 -0.492
ar.L2 -0.1923 7e-10 -2.75e+08 0.000 -0.192 -0.192
ar.L3 -0.0461 3.32e-10 -1.39e+08 0.000 -0.046 -0.046
ma.L1 -0.7077 2.73e-09 -2.59e+08 0.000 -0.708 -0.708
sigma2 8.99e-11 6.96e-11 1.291 0.197 -4.66e-11 2.26e-10
===================================================================================
Ljung-Box (L1) (Q): 55.51 Jarque-Bera (JB): 4268313.90
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 5.44
Prob(H) (two-sided): 0.00 Kurtosis: 359.56
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.36e+28. Standard errors may be unstable.
ARIMA order: (3, 3, 1)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.04775, saving model to LSTM8.h5
16/16 - 4s - loss: 1.4044 - val_loss: 0.0478 - lr: 0.0010 - 4s/epoch - 221ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.3789 - val_loss: 0.0488 - lr: 0.0010 - 109ms/epoch - 7ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.3506 - val_loss: 0.0499 - lr: 0.0010 - 95ms/epoch - 6ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.3191 - val_loss: 0.0511 - lr: 0.0010 - 86ms/epoch - 5ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2858 - val_loss: 0.0524 - lr: 0.0010 - 90ms/epoch - 6ms/step
Epoch 6/500
Epoch 00006: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00006: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2524 - val_loss: 0.0538 - lr: 0.0010 - 100ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2313 - val_loss: 0.0540 - lr: 1.0000e-04 - 89ms/epoch - 6ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2281 - val_loss: 0.0541 - lr: 1.0000e-04 - 98ms/epoch - 6ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2251 - val_loss: 0.0543 - lr: 1.0000e-04 - 90ms/epoch - 6ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2220 - val_loss: 0.0544 - lr: 1.0000e-04 - 100ms/epoch - 6ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00011: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2191 - val_loss: 0.0546 - lr: 1.0000e-04 - 98ms/epoch - 6ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2172 - val_loss: 0.0546 - lr: 1.0000e-05 - 89ms/epoch - 6ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2169 - val_loss: 0.0546 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2166 - val_loss: 0.0546 - lr: 1.0000e-05 - 106ms/epoch - 7ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2163 - val_loss: 0.0546 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00016: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2160 - val_loss: 0.0546 - lr: 1.0000e-05 - 112ms/epoch - 7ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2157 - val_loss: 0.0547 - lr: 1.0000e-05 - 86ms/epoch - 5ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2154 - val_loss: 0.0547 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2152 - val_loss: 0.0547 - lr: 1.0000e-05 - 93ms/epoch - 6ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2149 - val_loss: 0.0547 - lr: 1.0000e-05 - 86ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2146 - val_loss: 0.0547 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2143 - val_loss: 0.0547 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2140 - val_loss: 0.0548 - lr: 1.0000e-05 - 97ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2137 - val_loss: 0.0548 - lr: 1.0000e-05 - 92ms/epoch - 6ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2134 - val_loss: 0.0548 - lr: 1.0000e-05 - 89ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2132 - val_loss: 0.0548 - lr: 1.0000e-05 - 107ms/epoch - 7ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2129 - val_loss: 0.0548 - lr: 1.0000e-05 - 94ms/epoch - 6ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2126 - val_loss: 0.0548 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2123 - val_loss: 0.0549 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2120 - val_loss: 0.0549 - lr: 1.0000e-05 - 97ms/epoch - 6ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2117 - val_loss: 0.0549 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2114 - val_loss: 0.0549 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2112 - val_loss: 0.0549 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2109 - val_loss: 0.0549 - lr: 1.0000e-05 - 85ms/epoch - 5ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2106 - val_loss: 0.0550 - lr: 1.0000e-05 - 91ms/epoch - 6ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2103 - val_loss: 0.0550 - lr: 1.0000e-05 - 100ms/epoch - 6ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2100 - val_loss: 0.0550 - lr: 1.0000e-05 - 105ms/epoch - 7ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2098 - val_loss: 0.0550 - lr: 1.0000e-05 - 112ms/epoch - 7ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2095 - val_loss: 0.0550 - lr: 1.0000e-05 - 111ms/epoch - 7ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2092 - val_loss: 0.0550 - lr: 1.0000e-05 - 86ms/epoch - 5ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2089 - val_loss: 0.0551 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2086 - val_loss: 0.0551 - lr: 1.0000e-05 - 96ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2083 - val_loss: 0.0551 - lr: 1.0000e-05 - 96ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2081 - val_loss: 0.0551 - lr: 1.0000e-05 - 91ms/epoch - 6ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2078 - val_loss: 0.0551 - lr: 1.0000e-05 - 98ms/epoch - 6ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2075 - val_loss: 0.0551 - lr: 1.0000e-05 - 97ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2072 - val_loss: 0.0552 - lr: 1.0000e-05 - 105ms/epoch - 7ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2069 - val_loss: 0.0552 - lr: 1.0000e-05 - 104ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2067 - val_loss: 0.0552 - lr: 1.0000e-05 - 95ms/epoch - 6ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2064 - val_loss: 0.0552 - lr: 1.0000e-05 - 84ms/epoch - 5ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.04775
16/16 - 0s - loss: 1.2061 - val_loss: 0.0552 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step
Epoch 00051: early stopping
SMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 23.82392201920814
RMSE: 4.880975519218278
MAPE: 3.848681389687446
EMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 47.76% Accuracy
MSE: 36.51946703896653
RMSE: 6.043133875644865
MAPE: 4.745071953502077
WMA
WMA([input_arrays], [timeperiod=30])
Weighted Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
49
Working on WMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.780, Time=2.87 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.589, Time=4.52 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-16789.784, Time=11.92 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.589, Time=7.18 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16919.987, Time=9.64 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-14616.097, Time=11.83 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17225.955, Time=17.16 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-14562.589, Time=9.13 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-15582.364, Time=17.99 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-12043.670, Time=36.81 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 129.063 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8639.977
Date: Sun, 12 Dec 2021 AIC -17225.955
Time: 18:19:12 BIC -17099.302
Sample: 0 HQIC -17177.315
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -4.802e-09 4.51e-06 -0.001 0.999 -8.84e-06 8.83e-06
x2 -4.783e-09 4.5e-06 -0.001 0.999 -8.83e-06 8.82e-06
x3 -4.811e-09 4.51e-06 -0.001 0.999 -8.85e-06 8.84e-06
x4 1.0000 4.51e-06 2.22e+05 0.000 1.000 1.000
x5 -4.353e-09 4.3e-06 -0.001 0.999 -8.43e-06 8.42e-06
x6 -1.569e-08 7.54e-06 -0.002 0.998 -1.48e-05 1.48e-05
x7 -4.75e-09 4.49e-06 -0.001 0.999 -8.8e-06 8.79e-06
x8 -4.628e-09 4.43e-06 -0.001 0.999 -8.69e-06 8.69e-06
x9 -4.733e-10 1.16e-06 -0.000 1.000 -2.27e-06 2.27e-06
x10 -7.88e-10 1.8e-06 -0.000 1.000 -3.52e-06 3.52e-06
x11 -4.609e-09 4.42e-06 -0.001 0.999 -8.68e-06 8.67e-06
x12 -4.607e-09 4.42e-06 -0.001 0.999 -8.68e-06 8.67e-06
x13 -4.792e-09 4.51e-06 -0.001 0.999 -8.84e-06 8.83e-06
x14 -3.777e-08 1.24e-05 -0.003 0.998 -2.44e-05 2.44e-05
x15 -3.99e-09 4.12e-06 -0.001 0.999 -8.08e-06 8.07e-06
x16 -1.309e-08 7.41e-06 -0.002 0.999 -1.45e-05 1.45e-05
x17 -4.789e-09 4.51e-06 -0.001 0.999 -8.85e-06 8.84e-06
x18 -2.665e-10 9.77e-07 -0.000 1.000 -1.92e-06 1.92e-06
x19 -4.919e-09 4.56e-06 -0.001 0.999 -8.94e-06 8.93e-06
x20 -4e-10 9.58e-07 -0.000 1.000 -1.88e-06 1.88e-06
x21 -6.782e-08 1.67e-05 -0.004 0.997 -3.27e-05 3.26e-05
x22 -6.03e-08 1.58e-05 -0.004 0.997 -3.09e-05 3.08e-05
x23 -3.157e-08 1.14e-05 -0.003 0.998 -2.23e-05 2.23e-05
x24 -3.671e-08 1.23e-05 -0.003 0.998 -2.41e-05 2.41e-05
ma.L1 -1.3901 5.58e-10 -2.49e+09 0.000 -1.390 -1.390
ma.L2 0.4033 5.75e-10 7.02e+08 0.000 0.403 0.403
sigma2 7.525e-11 6.92e-11 1.088 0.277 -6.03e-11 2.11e-10
===================================================================================
Ljung-Box (L1) (Q): 69.18 Jarque-Bera (JB): 6366427.21
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 12.29
Prob(H) (two-sided): 0.00 Kurtosis: 437.97
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 2.29e+25. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.05123, saving model to LSTM8.h5
17/17 - 4s - loss: 1.4678 - val_loss: 0.0512 - lr: 0.0010 - 4s/epoch - 218ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.05123 to 0.05053, saving model to LSTM8.h5
17/17 - 0s - loss: 1.4292 - val_loss: 0.0505 - lr: 0.0010 - 113ms/epoch - 7ms/step
Epoch 3/500
Epoch 00003: val_loss improved from 0.05053 to 0.04967, saving model to LSTM8.h5
17/17 - 0s - loss: 1.3882 - val_loss: 0.0497 - lr: 0.0010 - 122ms/epoch - 7ms/step
Epoch 4/500
Epoch 00004: val_loss improved from 0.04967 to 0.04896, saving model to LSTM8.h5
17/17 - 0s - loss: 1.3479 - val_loss: 0.0490 - lr: 0.0010 - 121ms/epoch - 7ms/step
Epoch 5/500
Epoch 00005: val_loss improved from 0.04896 to 0.04876, saving model to LSTM8.h5
17/17 - 0s - loss: 1.3107 - val_loss: 0.0488 - lr: 0.0010 - 151ms/epoch - 9ms/step
Epoch 6/500
Epoch 00006: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.2726 - val_loss: 0.0492 - lr: 0.0010 - 119ms/epoch - 7ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.2304 - val_loss: 0.0503 - lr: 0.0010 - 110ms/epoch - 6ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.1883 - val_loss: 0.0519 - lr: 0.0010 - 119ms/epoch - 7ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.1513 - val_loss: 0.0538 - lr: 0.0010 - 102ms/epoch - 6ms/step
Epoch 10/500
Epoch 00010: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00010: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.1196 - val_loss: 0.0558 - lr: 0.0010 - 106ms/epoch - 6ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.1008 - val_loss: 0.0559 - lr: 1.0000e-04 - 99ms/epoch - 6ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0981 - val_loss: 0.0561 - lr: 1.0000e-04 - 109ms/epoch - 6ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0956 - val_loss: 0.0563 - lr: 1.0000e-04 - 118ms/epoch - 7ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0930 - val_loss: 0.0565 - lr: 1.0000e-04 - 119ms/epoch - 7ms/step
Epoch 15/500
Epoch 00015: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00015: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0906 - val_loss: 0.0567 - lr: 1.0000e-04 - 112ms/epoch - 7ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0890 - val_loss: 0.0567 - lr: 1.0000e-05 - 134ms/epoch - 8ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0887 - val_loss: 0.0567 - lr: 1.0000e-05 - 122ms/epoch - 7ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0885 - val_loss: 0.0567 - lr: 1.0000e-05 - 104ms/epoch - 6ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0882 - val_loss: 0.0567 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 20/500
Epoch 00020: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00020: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0880 - val_loss: 0.0568 - lr: 1.0000e-05 - 100ms/epoch - 6ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0878 - val_loss: 0.0568 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0875 - val_loss: 0.0568 - lr: 1.0000e-05 - 138ms/epoch - 8ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0873 - val_loss: 0.0568 - lr: 1.0000e-05 - 116ms/epoch - 7ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0870 - val_loss: 0.0568 - lr: 1.0000e-05 - 115ms/epoch - 7ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0868 - val_loss: 0.0569 - lr: 1.0000e-05 - 115ms/epoch - 7ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0866 - val_loss: 0.0569 - lr: 1.0000e-05 - 116ms/epoch - 7ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0863 - val_loss: 0.0569 - lr: 1.0000e-05 - 97ms/epoch - 6ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0861 - val_loss: 0.0569 - lr: 1.0000e-05 - 100ms/epoch - 6ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0858 - val_loss: 0.0570 - lr: 1.0000e-05 - 116ms/epoch - 7ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0856 - val_loss: 0.0570 - lr: 1.0000e-05 - 112ms/epoch - 7ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0854 - val_loss: 0.0570 - lr: 1.0000e-05 - 115ms/epoch - 7ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0851 - val_loss: 0.0570 - lr: 1.0000e-05 - 113ms/epoch - 7ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0849 - val_loss: 0.0570 - lr: 1.0000e-05 - 107ms/epoch - 6ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0846 - val_loss: 0.0571 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0844 - val_loss: 0.0571 - lr: 1.0000e-05 - 113ms/epoch - 7ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0842 - val_loss: 0.0571 - lr: 1.0000e-05 - 104ms/epoch - 6ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0839 - val_loss: 0.0571 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0837 - val_loss: 0.0571 - lr: 1.0000e-05 - 108ms/epoch - 6ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0834 - val_loss: 0.0572 - lr: 1.0000e-05 - 147ms/epoch - 9ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0832 - val_loss: 0.0572 - lr: 1.0000e-05 - 105ms/epoch - 6ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0830 - val_loss: 0.0572 - lr: 1.0000e-05 - 104ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0827 - val_loss: 0.0572 - lr: 1.0000e-05 - 108ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0825 - val_loss: 0.0573 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0822 - val_loss: 0.0573 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0820 - val_loss: 0.0573 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0818 - val_loss: 0.0573 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0815 - val_loss: 0.0573 - lr: 1.0000e-05 - 94ms/epoch - 6ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0813 - val_loss: 0.0574 - lr: 1.0000e-05 - 98ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0810 - val_loss: 0.0574 - lr: 1.0000e-05 - 103ms/epoch - 6ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0808 - val_loss: 0.0574 - lr: 1.0000e-05 - 102ms/epoch - 6ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0806 - val_loss: 0.0574 - lr: 1.0000e-05 - 105ms/epoch - 6ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0803 - val_loss: 0.0575 - lr: 1.0000e-05 - 128ms/epoch - 8ms/step
Epoch 53/500
Epoch 00053: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0801 - val_loss: 0.0575 - lr: 1.0000e-05 - 106ms/epoch - 6ms/step
Epoch 54/500
Epoch 00054: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0799 - val_loss: 0.0575 - lr: 1.0000e-05 - 104ms/epoch - 6ms/step
Epoch 55/500
Epoch 00055: val_loss did not improve from 0.04876
17/17 - 0s - loss: 1.0796 - val_loss: 0.0575 - lr: 1.0000e-05 - 99ms/epoch - 6ms/step
Epoch 00055: early stopping
SMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 23.82392201920814
RMSE: 4.880975519218278
MAPE: 3.848681389687446
EMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 47.76% Accuracy
MSE: 36.51946703896653
RMSE: 6.043133875644865
MAPE: 4.745071953502077
WMA
Prediction vs Close: 53.73% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 43.8622174139386
RMSE: 6.622855684214975
MAPE: 5.251112001385055
DEMA
DEMA([input_arrays], [timeperiod=30])
Double Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
89
Working on DEMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.785, Time=3.17 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.588, Time=4.72 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-15575.689, Time=9.20 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.588, Time=7.35 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16714.796, Time=9.20 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-15610.140, Time=11.04 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17225.835, Time=22.73 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-14562.588, Time=9.27 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-16751.951, Time=21.21 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-11788.089, Time=32.23 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 130.137 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8639.917
Date: Sun, 12 Dec 2021 AIC -17225.835
Time: 18:25:56 BIC -17099.182
Sample: 0 HQIC -17177.195
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -5.894e-09 3.61e-06 -0.002 0.999 -7.09e-06 7.08e-06
x2 -5.93e-09 3.63e-06 -0.002 0.999 -7.11e-06 7.1e-06
x3 -5.905e-09 3.62e-06 -0.002 0.999 -7.1e-06 7.09e-06
x4 1.0000 3.62e-06 2.76e+05 0.000 1.000 1.000
x5 -5.457e-09 3.48e-06 -0.002 0.999 -6.83e-06 6.82e-06
x6 -3.019e-08 7.72e-06 -0.004 0.997 -1.52e-05 1.51e-05
x7 -5.87e-09 3.61e-06 -0.002 0.999 -7.08e-06 7.07e-06
x8 -5.809e-09 3.59e-06 -0.002 0.999 -7.05e-06 7.04e-06
x9 -9.293e-11 9.83e-08 -0.001 0.999 -1.93e-07 1.93e-07
x10 -2.793e-09 2.47e-06 -0.001 0.999 -4.84e-06 4.84e-06
x11 -6.095e-09 3.68e-06 -0.002 0.999 -7.21e-06 7.2e-06
x12 -5.478e-09 3.49e-06 -0.002 0.999 -6.85e-06 6.84e-06
x13 -5.91e-09 3.62e-06 -0.002 0.999 -7.1e-06 7.09e-06
x14 -4.085e-08 9.35e-06 -0.004 0.997 -1.84e-05 1.83e-05
x15 -5.93e-09 3.63e-06 -0.002 0.999 -7.12e-06 7.11e-06
x16 -1.618e-09 1.92e-06 -0.001 0.999 -3.76e-06 3.75e-06
x17 -5.076e-09 3.37e-06 -0.002 0.999 -6.6e-06 6.59e-06
x18 -1.377e-08 5.5e-06 -0.003 0.998 -1.08e-05 1.08e-05
x19 -6.135e-09 3.69e-06 -0.002 0.999 -7.23e-06 7.22e-06
x20 -1.018e-08 4.43e-06 -0.002 0.998 -8.68e-06 8.66e-06
x21 -6.911e-08 1.21e-05 -0.006 0.995 -2.39e-05 2.37e-05
x22 -5.656e-08 1.1e-05 -0.005 0.996 -2.16e-05 2.15e-05
x23 -5.355e-08 1.07e-05 -0.005 0.996 -2.1e-05 2.09e-05
x24 -3.636e-08 8.85e-06 -0.004 0.997 -1.74e-05 1.73e-05
ma.L1 -1.3899 4.86e-11 -2.86e+10 0.000 -1.390 -1.390
ma.L2 0.4032 4.6e-11 8.76e+09 0.000 0.403 0.403
sigma2 7.526e-11 6.92e-11 1.088 0.277 -6.03e-11 2.11e-10
===================================================================================
Ljung-Box (L1) (Q): 69.65 Jarque-Bera (JB): 6422892.15
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 12.42
Prob(H) (two-sided): 0.00 Kurtosis: 439.89
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 2.74e+29. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.04977, saving model to LSTM8.h5
10/10 - 4s - loss: 1.4372 - val_loss: 0.0498 - lr: 0.0010 - 4s/epoch - 418ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.4260 - val_loss: 0.0503 - lr: 0.0010 - 72ms/epoch - 7ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.4155 - val_loss: 0.0510 - lr: 0.0010 - 66ms/epoch - 7ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.4050 - val_loss: 0.0516 - lr: 0.0010 - 76ms/epoch - 8ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3941 - val_loss: 0.0523 - lr: 0.0010 - 73ms/epoch - 7ms/step
Epoch 6/500
Epoch 00006: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00006: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3829 - val_loss: 0.0530 - lr: 0.0010 - 79ms/epoch - 8ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3749 - val_loss: 0.0531 - lr: 1.0000e-04 - 80ms/epoch - 8ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3737 - val_loss: 0.0532 - lr: 1.0000e-04 - 65ms/epoch - 6ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3725 - val_loss: 0.0533 - lr: 1.0000e-04 - 84ms/epoch - 8ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3714 - val_loss: 0.0533 - lr: 1.0000e-04 - 68ms/epoch - 7ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00011: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3702 - val_loss: 0.0534 - lr: 1.0000e-04 - 79ms/epoch - 8ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3694 - val_loss: 0.0534 - lr: 1.0000e-05 - 96ms/epoch - 10ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3693 - val_loss: 0.0534 - lr: 1.0000e-05 - 85ms/epoch - 9ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3692 - val_loss: 0.0534 - lr: 1.0000e-05 - 85ms/epoch - 9ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3691 - val_loss: 0.0534 - lr: 1.0000e-05 - 73ms/epoch - 7ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00016: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3689 - val_loss: 0.0534 - lr: 1.0000e-05 - 84ms/epoch - 8ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3688 - val_loss: 0.0534 - lr: 1.0000e-05 - 75ms/epoch - 7ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3687 - val_loss: 0.0535 - lr: 1.0000e-05 - 74ms/epoch - 7ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3686 - val_loss: 0.0535 - lr: 1.0000e-05 - 67ms/epoch - 7ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3685 - val_loss: 0.0535 - lr: 1.0000e-05 - 73ms/epoch - 7ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3684 - val_loss: 0.0535 - lr: 1.0000e-05 - 65ms/epoch - 6ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3683 - val_loss: 0.0535 - lr: 1.0000e-05 - 84ms/epoch - 8ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3681 - val_loss: 0.0535 - lr: 1.0000e-05 - 65ms/epoch - 7ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3680 - val_loss: 0.0535 - lr: 1.0000e-05 - 71ms/epoch - 7ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3679 - val_loss: 0.0535 - lr: 1.0000e-05 - 89ms/epoch - 9ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3678 - val_loss: 0.0535 - lr: 1.0000e-05 - 91ms/epoch - 9ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3677 - val_loss: 0.0535 - lr: 1.0000e-05 - 80ms/epoch - 8ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3676 - val_loss: 0.0535 - lr: 1.0000e-05 - 80ms/epoch - 8ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3675 - val_loss: 0.0535 - lr: 1.0000e-05 - 88ms/epoch - 9ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3674 - val_loss: 0.0535 - lr: 1.0000e-05 - 88ms/epoch - 9ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3672 - val_loss: 0.0536 - lr: 1.0000e-05 - 87ms/epoch - 9ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3671 - val_loss: 0.0536 - lr: 1.0000e-05 - 71ms/epoch - 7ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3670 - val_loss: 0.0536 - lr: 1.0000e-05 - 86ms/epoch - 9ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3669 - val_loss: 0.0536 - lr: 1.0000e-05 - 77ms/epoch - 8ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3668 - val_loss: 0.0536 - lr: 1.0000e-05 - 93ms/epoch - 9ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3667 - val_loss: 0.0536 - lr: 1.0000e-05 - 95ms/epoch - 9ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3666 - val_loss: 0.0536 - lr: 1.0000e-05 - 81ms/epoch - 8ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3664 - val_loss: 0.0536 - lr: 1.0000e-05 - 70ms/epoch - 7ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3663 - val_loss: 0.0536 - lr: 1.0000e-05 - 89ms/epoch - 9ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3662 - val_loss: 0.0536 - lr: 1.0000e-05 - 81ms/epoch - 8ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3661 - val_loss: 0.0536 - lr: 1.0000e-05 - 80ms/epoch - 8ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3660 - val_loss: 0.0536 - lr: 1.0000e-05 - 76ms/epoch - 8ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3659 - val_loss: 0.0536 - lr: 1.0000e-05 - 70ms/epoch - 7ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3658 - val_loss: 0.0536 - lr: 1.0000e-05 - 87ms/epoch - 9ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3657 - val_loss: 0.0537 - lr: 1.0000e-05 - 71ms/epoch - 7ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3655 - val_loss: 0.0537 - lr: 1.0000e-05 - 77ms/epoch - 8ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3654 - val_loss: 0.0537 - lr: 1.0000e-05 - 70ms/epoch - 7ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3653 - val_loss: 0.0537 - lr: 1.0000e-05 - 72ms/epoch - 7ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3652 - val_loss: 0.0537 - lr: 1.0000e-05 - 89ms/epoch - 9ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3651 - val_loss: 0.0537 - lr: 1.0000e-05 - 73ms/epoch - 7ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.04977
10/10 - 0s - loss: 1.3650 - val_loss: 0.0537 - lr: 1.0000e-05 - 87ms/epoch - 9ms/step
Epoch 00051: early stopping
SMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 23.82392201920814
RMSE: 4.880975519218278
MAPE: 3.848681389687446
EMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 47.76% Accuracy
MSE: 36.51946703896653
RMSE: 6.043133875644865
MAPE: 4.745071953502077
WMA
Prediction vs Close: 53.73% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 43.8622174139386
RMSE: 6.622855684214975
MAPE: 5.251112001385055
DEMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 142.0772798190819
RMSE: 11.91961743593652
MAPE: 10.62322903609799
KAMA
KAMA([input_arrays], [timeperiod=30])
Kaufman Adaptive Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
18
Working on KAMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16921.943, Time=10.33 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.592, Time=4.80 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-16797.275, Time=9.62 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.592, Time=7.38 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16996.465, Time=3.46 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=-16999.509, Time=3.29 sec
ARIMA(3,3,0)(0,0,0)[0] : AIC=-17171.315, Time=6.42 sec
ARIMA(3,3,1)(0,0,0)[0] : AIC=-16994.523, Time=4.03 sec
ARIMA(3,3,0)(0,0,0)[0] intercept : AIC=-15518.026, Time=31.80 sec
Best model: ARIMA(3,3,0)(0,0,0)[0]
Total fit time: 81.148 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(3, 3, 0) Log Likelihood 8613.658
Date: Sun, 12 Dec 2021 AIC -17171.315
Time: 18:32:05 BIC -17039.972
Sample: 0 HQIC -17120.874
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -5.14e-10 7.6e-05 -6.76e-06 1.000 -0.000 0.000
x2 -5.041e-10 7.52e-05 -6.7e-06 1.000 -0.000 0.000
x3 -4.834e-10 7.38e-05 -6.55e-06 1.000 -0.000 0.000
x4 1.0000 7.46e-05 1.34e+04 0.000 1.000 1.000
x5 -4.462e-10 7.09e-05 -6.29e-06 1.000 -0.000 0.000
x6 -3.064e-09 0.000 -1.84e-05 1.000 -0.000 0.000
x7 -4.751e-10 7.35e-05 -6.46e-06 1.000 -0.000 0.000
x8 -4.628e-10 7.28e-05 -6.36e-06 1.000 -0.000 0.000
x9 -9.21e-11 9.37e-06 -9.83e-06 1.000 -1.84e-05 1.84e-05
x10 -2.165e-10 3.1e-05 -6.98e-06 1.000 -6.08e-05 6.08e-05
x11 -4.665e-10 7.28e-05 -6.41e-06 1.000 -0.000 0.000
x12 -4.62e-10 7.23e-05 -6.39e-06 1.000 -0.000 0.000
x13 -4.906e-10 7.43e-05 -6.6e-06 1.000 -0.000 0.000
x14 -3.985e-09 0.000 -1.87e-05 1.000 -0.000 0.000
x15 -4.897e-10 7.48e-05 -6.55e-06 1.000 -0.000 0.000
x16 -7.327e-10 9.24e-05 -7.93e-06 1.000 -0.000 0.000
x17 -4.173e-10 6.93e-05 -6.02e-06 1.000 -0.000 0.000
x18 -3.397e-10 6.02e-05 -5.64e-06 1.000 -0.000 0.000
x19 -6.012e-10 8.3e-05 -7.25e-06 1.000 -0.000 0.000
x20 -9.09e-10 0.000 -9.05e-06 1.000 -0.000 0.000
x21 -6.188e-09 0.000 -2.32e-05 1.000 -0.001 0.001
x22 -1.992e-09 0.000 -1.33e-05 1.000 -0.000 0.000
x23 -3.669e-09 0.000 -1.79e-05 1.000 -0.000 0.000
x24 -1.065e-09 0.000 -1.01e-05 1.000 -0.000 0.000
ar.L1 -1.2073 5.73e-10 -2.11e+09 0.000 -1.207 -1.207
ar.L2 -0.9083 5.93e-10 -1.53e+09 0.000 -0.908 -0.908
ar.L3 -0.4033 5.84e-10 -6.91e+08 0.000 -0.403 -0.403
sigma2 8.06e-11 6.94e-11 1.162 0.245 -5.54e-11 2.17e-10
===================================================================================
Ljung-Box (L1) (Q): 13.77 Jarque-Bera (JB): 2436796.68
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: 4.07
Prob(H) (two-sided): 0.00 Kurtosis: 272.41
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.23e+28. Standard errors may be unstable.
ARIMA order: (3, 3, 0)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.05006, saving model to LSTM8.h5
45/45 - 4s - loss: 1.4094 - val_loss: 0.0501 - lr: 0.0010 - 4s/epoch - 86ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.05006
45/45 - 0s - loss: 1.3333 - val_loss: 0.0534 - lr: 0.0010 - 263ms/epoch - 6ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.05006
45/45 - 0s - loss: 1.2221 - val_loss: 0.0572 - lr: 0.0010 - 253ms/epoch - 6ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.05006
45/45 - 0s - loss: 1.1381 - val_loss: 0.0608 - lr: 0.0010 - 279ms/epoch - 6ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.05006
45/45 - 0s - loss: 1.0759 - val_loss: 0.0644 - lr: 0.0010 - 254ms/epoch - 6ms/step
Epoch 6/500
Epoch 00006: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00006: val_loss did not improve from 0.05006
45/45 - 0s - loss: 1.0236 - val_loss: 0.0684 - lr: 0.0010 - 269ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9942 - val_loss: 0.0689 - lr: 1.0000e-04 - 255ms/epoch - 6ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9898 - val_loss: 0.0693 - lr: 1.0000e-04 - 264ms/epoch - 6ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9855 - val_loss: 0.0698 - lr: 1.0000e-04 - 218ms/epoch - 5ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9812 - val_loss: 0.0703 - lr: 1.0000e-04 - 231ms/epoch - 5ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00011: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9770 - val_loss: 0.0707 - lr: 1.0000e-04 - 300ms/epoch - 7ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9744 - val_loss: 0.0708 - lr: 1.0000e-05 - 244ms/epoch - 5ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9740 - val_loss: 0.0708 - lr: 1.0000e-05 - 232ms/epoch - 5ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9736 - val_loss: 0.0709 - lr: 1.0000e-05 - 252ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9731 - val_loss: 0.0710 - lr: 1.0000e-05 - 267ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00016: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9727 - val_loss: 0.0710 - lr: 1.0000e-05 - 227ms/epoch - 5ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9723 - val_loss: 0.0711 - lr: 1.0000e-05 - 222ms/epoch - 5ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9719 - val_loss: 0.0711 - lr: 1.0000e-05 - 234ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9715 - val_loss: 0.0712 - lr: 1.0000e-05 - 270ms/epoch - 6ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9710 - val_loss: 0.0712 - lr: 1.0000e-05 - 266ms/epoch - 6ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9706 - val_loss: 0.0713 - lr: 1.0000e-05 - 281ms/epoch - 6ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9702 - val_loss: 0.0713 - lr: 1.0000e-05 - 249ms/epoch - 6ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9698 - val_loss: 0.0714 - lr: 1.0000e-05 - 260ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9694 - val_loss: 0.0715 - lr: 1.0000e-05 - 236ms/epoch - 5ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9689 - val_loss: 0.0715 - lr: 1.0000e-05 - 256ms/epoch - 6ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9685 - val_loss: 0.0716 - lr: 1.0000e-05 - 262ms/epoch - 6ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9681 - val_loss: 0.0716 - lr: 1.0000e-05 - 275ms/epoch - 6ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9677 - val_loss: 0.0717 - lr: 1.0000e-05 - 257ms/epoch - 6ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9672 - val_loss: 0.0718 - lr: 1.0000e-05 - 223ms/epoch - 5ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9668 - val_loss: 0.0718 - lr: 1.0000e-05 - 265ms/epoch - 6ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9664 - val_loss: 0.0719 - lr: 1.0000e-05 - 276ms/epoch - 6ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9660 - val_loss: 0.0719 - lr: 1.0000e-05 - 269ms/epoch - 6ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9655 - val_loss: 0.0720 - lr: 1.0000e-05 - 242ms/epoch - 5ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9651 - val_loss: 0.0721 - lr: 1.0000e-05 - 253ms/epoch - 6ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9647 - val_loss: 0.0721 - lr: 1.0000e-05 - 285ms/epoch - 6ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9643 - val_loss: 0.0722 - lr: 1.0000e-05 - 272ms/epoch - 6ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9638 - val_loss: 0.0723 - lr: 1.0000e-05 - 245ms/epoch - 5ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9634 - val_loss: 0.0723 - lr: 1.0000e-05 - 251ms/epoch - 6ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9630 - val_loss: 0.0724 - lr: 1.0000e-05 - 284ms/epoch - 6ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9626 - val_loss: 0.0725 - lr: 1.0000e-05 - 256ms/epoch - 6ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9621 - val_loss: 0.0725 - lr: 1.0000e-05 - 239ms/epoch - 5ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9617 - val_loss: 0.0726 - lr: 1.0000e-05 - 271ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9613 - val_loss: 0.0727 - lr: 1.0000e-05 - 277ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9609 - val_loss: 0.0727 - lr: 1.0000e-05 - 282ms/epoch - 6ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9605 - val_loss: 0.0728 - lr: 1.0000e-05 - 233ms/epoch - 5ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9600 - val_loss: 0.0729 - lr: 1.0000e-05 - 269ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9596 - val_loss: 0.0729 - lr: 1.0000e-05 - 245ms/epoch - 5ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9592 - val_loss: 0.0730 - lr: 1.0000e-05 - 238ms/epoch - 5ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9588 - val_loss: 0.0731 - lr: 1.0000e-05 - 278ms/epoch - 6ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9583 - val_loss: 0.0731 - lr: 1.0000e-05 - 297ms/epoch - 7ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.05006
45/45 - 0s - loss: 0.9579 - val_loss: 0.0732 - lr: 1.0000e-05 - 236ms/epoch - 5ms/step
Epoch 00051: early stopping
SMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 23.82392201920814
RMSE: 4.880975519218278
MAPE: 3.848681389687446
EMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 47.76% Accuracy
MSE: 36.51946703896653
RMSE: 6.043133875644865
MAPE: 4.745071953502077
WMA
Prediction vs Close: 53.73% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 43.8622174139386
RMSE: 6.622855684214975
MAPE: 5.251112001385055
DEMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 142.0772798190819
RMSE: 11.91961743593652
MAPE: 10.62322903609799
KAMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 50.0% Accuracy
MSE: 22.038946216129663
RMSE: 4.694565604625168
MAPE: 3.7548184287573756
MIDPOINT
MIDPOINT([input_arrays], [timeperiod=14])
MidPoint over period (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 14
Outputs:
real
14
Working on MIDPOINT predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16999.768, Time=3.13 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14568.591, Time=4.75 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-15581.065, Time=9.16 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-14566.591, Time=7.15 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16536.628, Time=9.56 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-13971.493, Time=10.03 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17226.044, Time=22.66 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-14562.591, Time=9.61 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-16754.945, Time=20.07 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-15001.855, Time=20.68 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 116.804 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8640.022
Date: Sun, 12 Dec 2021 AIC -17226.044
Time: 18:35:59 BIC -17099.391
Sample: 0 HQIC -17177.404
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -5.031e-09 1.06e-05 -0.000 1.000 -2.08e-05 2.08e-05
x2 -4.99e-09 8.12e-06 -0.001 1.000 -1.59e-05 1.59e-05
x3 -5.114e-09 1.38e-05 -0.000 1.000 -2.7e-05 2.7e-05
x4 1.0000 8.91e-06 1.12e+05 0.000 1.000 1.000
x5 -4.55e-09 8.2e-06 -0.001 1.000 -1.61e-05 1.61e-05
x6 -9.992e-08 0.001 -0.000 1.000 -0.002 0.002
x7 -4.607e-09 1.97e-05 -0.000 1.000 -3.86e-05 3.86e-05
x8 -4.591e-09 1.77e-05 -0.000 1.000 -3.48e-05 3.48e-05
x9 -2.538e-09 1.13e-05 -0.000 1.000 -2.21e-05 2.21e-05
x10 -4.315e-09 6.08e-06 -0.001 0.999 -1.19e-05 1.19e-05
x11 -4.545e-09 1.62e-05 -0.000 1.000 -3.18e-05 3.18e-05
x12 -4.701e-09 1.97e-05 -0.000 1.000 -3.87e-05 3.87e-05
x13 -4.823e-09 1.18e-05 -0.000 1.000 -2.3e-05 2.3e-05
x14 -4.08e-08 4.99e-05 -0.001 0.999 -9.79e-05 9.78e-05
x15 -5.557e-09 2.03e-05 -0.000 1.000 -3.99e-05 3.99e-05
x16 -3.541e-09 1.3e-05 -0.000 1.000 -2.55e-05 2.55e-05
x17 -3.463e-09 1.51e-05 -0.000 1.000 -2.97e-05 2.97e-05
x18 -1.534e-08 4e-05 -0.000 1.000 -7.85e-05 7.85e-05
x19 -6.118e-09 2.07e-05 -0.000 1.000 -4.05e-05 4.05e-05
x20 -1.581e-08 3.38e-05 -0.000 1.000 -6.62e-05 6.61e-05
x21 -5.505e-08 5.6e-05 -0.001 0.999 -0.000 0.000
x22 -2.936e-08 4.55e-05 -0.001 0.999 -8.92e-05 8.92e-05
x23 -3.882e-08 4.89e-05 -0.001 0.999 -9.58e-05 9.57e-05
x24 -2.099e-08 4.87e-05 -0.000 1.000 -9.54e-05 9.54e-05
ma.L1 -1.3900 1.23e-07 -1.13e+07 0.000 -1.390 -1.390
ma.L2 0.4044 1.43e-07 2.82e+06 0.000 0.404 0.404
sigma2 7.525e-11 7.22e-11 1.042 0.297 -6.63e-11 2.17e-10
===================================================================================
Ljung-Box (L1) (Q): 55.84 Jarque-Bera (JB): 1335305.59
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.09 Skew: 5.74
Prob(H) (two-sided): 0.00 Kurtosis: 202.19
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.77e+23. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.05306, saving model to LSTM8.h5
58/58 - 4s - loss: 1.4273 - val_loss: 0.0531 - lr: 0.0010 - 4s/epoch - 67ms/step
Epoch 2/500
Epoch 00002: val_loss improved from 0.05306 to 0.05231, saving model to LSTM8.h5
58/58 - 0s - loss: 1.3303 - val_loss: 0.0523 - lr: 0.0010 - 293ms/epoch - 5ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.05231
58/58 - 0s - loss: 1.1533 - val_loss: 0.0582 - lr: 0.0010 - 306ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.9755 - val_loss: 0.0665 - lr: 0.0010 - 310ms/epoch - 5ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.8640 - val_loss: 0.0698 - lr: 0.0010 - 299ms/epoch - 5ms/step
Epoch 6/500
Epoch 00006: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7991 - val_loss: 0.0731 - lr: 0.0010 - 333ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00007: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7638 - val_loss: 0.0768 - lr: 0.0010 - 310ms/epoch - 5ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7472 - val_loss: 0.0771 - lr: 1.0000e-04 - 303ms/epoch - 5ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7448 - val_loss: 0.0775 - lr: 1.0000e-04 - 294ms/epoch - 5ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7424 - val_loss: 0.0780 - lr: 1.0000e-04 - 304ms/epoch - 5ms/step
Epoch 11/500
Epoch 00011: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7400 - val_loss: 0.0784 - lr: 1.0000e-04 - 310ms/epoch - 5ms/step
Epoch 12/500
Epoch 00012: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00012: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7376 - val_loss: 0.0788 - lr: 1.0000e-04 - 298ms/epoch - 5ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7362 - val_loss: 0.0789 - lr: 1.0000e-05 - 302ms/epoch - 5ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7359 - val_loss: 0.0789 - lr: 1.0000e-05 - 299ms/epoch - 5ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7357 - val_loss: 0.0790 - lr: 1.0000e-05 - 332ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7355 - val_loss: 0.0790 - lr: 1.0000e-05 - 300ms/epoch - 5ms/step
Epoch 17/500
Epoch 00017: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00017: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7352 - val_loss: 0.0791 - lr: 1.0000e-05 - 304ms/epoch - 5ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7350 - val_loss: 0.0792 - lr: 1.0000e-05 - 283ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7347 - val_loss: 0.0792 - lr: 1.0000e-05 - 297ms/epoch - 5ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7345 - val_loss: 0.0793 - lr: 1.0000e-05 - 314ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7342 - val_loss: 0.0793 - lr: 1.0000e-05 - 321ms/epoch - 6ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7339 - val_loss: 0.0794 - lr: 1.0000e-05 - 304ms/epoch - 5ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7337 - val_loss: 0.0794 - lr: 1.0000e-05 - 332ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7334 - val_loss: 0.0795 - lr: 1.0000e-05 - 326ms/epoch - 6ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7331 - val_loss: 0.0796 - lr: 1.0000e-05 - 285ms/epoch - 5ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7329 - val_loss: 0.0796 - lr: 1.0000e-05 - 377ms/epoch - 7ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7326 - val_loss: 0.0797 - lr: 1.0000e-05 - 296ms/epoch - 5ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7323 - val_loss: 0.0798 - lr: 1.0000e-05 - 289ms/epoch - 5ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7321 - val_loss: 0.0798 - lr: 1.0000e-05 - 319ms/epoch - 5ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7318 - val_loss: 0.0799 - lr: 1.0000e-05 - 293ms/epoch - 5ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7315 - val_loss: 0.0800 - lr: 1.0000e-05 - 302ms/epoch - 5ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7313 - val_loss: 0.0801 - lr: 1.0000e-05 - 294ms/epoch - 5ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7310 - val_loss: 0.0801 - lr: 1.0000e-05 - 304ms/epoch - 5ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7307 - val_loss: 0.0802 - lr: 1.0000e-05 - 292ms/epoch - 5ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7304 - val_loss: 0.0803 - lr: 1.0000e-05 - 305ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7301 - val_loss: 0.0804 - lr: 1.0000e-05 - 391ms/epoch - 7ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7299 - val_loss: 0.0805 - lr: 1.0000e-05 - 305ms/epoch - 5ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7296 - val_loss: 0.0805 - lr: 1.0000e-05 - 301ms/epoch - 5ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7293 - val_loss: 0.0806 - lr: 1.0000e-05 - 320ms/epoch - 6ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7290 - val_loss: 0.0807 - lr: 1.0000e-05 - 315ms/epoch - 5ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7288 - val_loss: 0.0808 - lr: 1.0000e-05 - 314ms/epoch - 5ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7285 - val_loss: 0.0809 - lr: 1.0000e-05 - 323ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7282 - val_loss: 0.0810 - lr: 1.0000e-05 - 324ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7279 - val_loss: 0.0811 - lr: 1.0000e-05 - 305ms/epoch - 5ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7276 - val_loss: 0.0811 - lr: 1.0000e-05 - 295ms/epoch - 5ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7274 - val_loss: 0.0812 - lr: 1.0000e-05 - 303ms/epoch - 5ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7271 - val_loss: 0.0813 - lr: 1.0000e-05 - 305ms/epoch - 5ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7268 - val_loss: 0.0814 - lr: 1.0000e-05 - 328ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7265 - val_loss: 0.0815 - lr: 1.0000e-05 - 369ms/epoch - 6ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7262 - val_loss: 0.0816 - lr: 1.0000e-05 - 301ms/epoch - 5ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7260 - val_loss: 0.0817 - lr: 1.0000e-05 - 330ms/epoch - 6ms/step
Epoch 52/500
Epoch 00052: val_loss did not improve from 0.05231
58/58 - 0s - loss: 0.7257 - val_loss: 0.0818 - lr: 1.0000e-05 - 306ms/epoch - 5ms/step
Epoch 00052: early stopping
SMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 23.82392201920814
RMSE: 4.880975519218278
MAPE: 3.848681389687446
EMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 47.76% Accuracy
MSE: 36.51946703896653
RMSE: 6.043133875644865
MAPE: 4.745071953502077
WMA
Prediction vs Close: 53.73% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 43.8622174139386
RMSE: 6.622855684214975
MAPE: 5.251112001385055
DEMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 142.0772798190819
RMSE: 11.91961743593652
MAPE: 10.62322903609799
KAMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 50.0% Accuracy
MSE: 22.038946216129663
RMSE: 4.694565604625168
MAPE: 3.7548184287573756
MIDPOINT
Prediction vs Close: 51.87% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 18.4805779796863
RMSE: 4.298904276636815
MAPE: 3.422385888494848
T3
T3([input_arrays], [timeperiod=5], [vfactor=0.7])
Triple Exponential Moving Average (T3) (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 5
vfactor: 0.7
Outputs:
real
19
Working on T3 predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-17000.569, Time=3.53 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-15576.554, Time=5.81 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-16078.305, Time=8.31 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-15574.554, Time=9.30 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-16998.627, Time=3.30 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-16429.916, Time=12.90 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-17000.664, Time=3.34 sec
ARIMA(0,3,3)(0,0,0)[0] : AIC=-15700.026, Time=11.65 sec
ARIMA(1,3,3)(0,0,0)[0] : AIC=-15704.282, Time=15.57 sec
ARIMA(0,3,2)(0,0,0)[0] intercept : AIC=-16998.664, Time=3.35 sec
Best model: ARIMA(0,3,2)(0,0,0)[0]
Total fit time: 77.090 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(0, 3, 2) Log Likelihood 8527.332
Date: Sun, 12 Dec 2021 AIC -17000.664
Time: 18:41:53 BIC -16874.011
Sample: 0 HQIC -16952.024
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 8.378e-14 2.16e-06 3.89e-08 1.000 -4.23e-06 4.23e-06
x2 7.457e-14 2.15e-06 3.47e-08 1.000 -4.22e-06 4.22e-06
x3 2.279e-14 2.16e-06 1.05e-08 1.000 -4.24e-06 4.24e-06
x4 1.0000 2.16e-06 4.63e+05 0.000 1.000 1.000
x5 1.211e-12 2.07e-06 5.86e-07 1.000 -4.05e-06 4.05e-06
x6 3.146e-15 2.67e-06 1.18e-09 1.000 -5.23e-06 5.23e-06
x7 1.593e-13 2.15e-06 7.41e-08 1.000 -4.21e-06 4.21e-06
x8 -0.0001 2.1e-06 -48.778 0.000 -0.000 -9.82e-05
x9 5.141e-14 6.35e-07 8.1e-08 1.000 -1.24e-06 1.24e-06
x10 -6.174e-05 1.34e-06 -45.995 0.000 -6.44e-05 -5.91e-05
x11 0.0003 2.15e-06 148.354 0.000 0.000 0.000
x12 -0.0002 2.02e-06 -93.730 0.000 -0.000 -0.000
x13 1.967e-14 2.16e-06 9.1e-09 1.000 -4.23e-06 4.23e-06
x14 -1.297e-14 5.65e-06 -2.29e-09 1.000 -1.11e-05 1.11e-05
x15 -3.18e-12 1.82e-06 -1.75e-06 1.000 -3.57e-06 3.57e-06
x16 -1.426e-12 4.51e-06 -3.16e-07 1.000 -8.84e-06 8.84e-06
x17 7.474e-13 2.37e-06 3.16e-07 1.000 -4.64e-06 4.64e-06
x18 -2.92e-13 2.9e-06 -1.01e-07 1.000 -5.68e-06 5.68e-06
x19 -4.211e-14 1.89e-06 -2.22e-08 1.000 -3.71e-06 3.71e-06
x20 -1.515e-13 1.2e-06 -1.26e-07 1.000 -2.36e-06 2.36e-06
x21 6.555e-13 6.37e-06 1.03e-07 1.000 -1.25e-05 1.25e-05
x22 1.212e-14 6.19e-06 1.96e-09 1.000 -1.21e-05 1.21e-05
x23 -3.877e-13 3.76e-06 -1.03e-07 1.000 -7.38e-06 7.38e-06
x24 8.127e-15 4.01e-06 2.03e-09 1.000 -7.86e-06 7.86e-06
ma.L1 -1.3370 3.84e-12 -3.48e+11 0.000 -1.337 -1.337
ma.L2 0.4289 1.65e-12 2.6e+11 0.000 0.429 0.429
sigma2 1e-10 6.99e-11 1.430 0.153 -3.71e-11 2.37e-10
===================================================================================
Ljung-Box (L1) (Q): 4.57 Jarque-Bera (JB): 3228712.87
Prob(Q): 0.03 Prob(JB): 0.00
Heteroskedasticity (H): 0.12 Skew: -9.87
Prob(H) (two-sided): 0.00 Kurtosis: 312.63
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 1.57e+30. Standard errors may be unstable.
ARIMA order: (0, 3, 2)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.04608, saving model to LSTM8.h5
43/43 - 4s - loss: 1.3504 - val_loss: 0.0461 - lr: 0.0010 - 4s/epoch - 102ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.04608
43/43 - 0s - loss: 1.2322 - val_loss: 0.0478 - lr: 0.0010 - 225ms/epoch - 5ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.04608
43/43 - 0s - loss: 1.1310 - val_loss: 0.0496 - lr: 0.0010 - 248ms/epoch - 6ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.04608
43/43 - 0s - loss: 1.0501 - val_loss: 0.0517 - lr: 0.0010 - 243ms/epoch - 6ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.9843 - val_loss: 0.0543 - lr: 0.0010 - 252ms/epoch - 6ms/step
Epoch 6/500
Epoch 00006: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00006: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.9252 - val_loss: 0.0580 - lr: 0.0010 - 225ms/epoch - 5ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8901 - val_loss: 0.0584 - lr: 1.0000e-04 - 243ms/epoch - 6ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8856 - val_loss: 0.0588 - lr: 1.0000e-04 - 260ms/epoch - 6ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8812 - val_loss: 0.0592 - lr: 1.0000e-04 - 278ms/epoch - 6ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8770 - val_loss: 0.0596 - lr: 1.0000e-04 - 226ms/epoch - 5ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00011: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8730 - val_loss: 0.0600 - lr: 1.0000e-04 - 228ms/epoch - 5ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8706 - val_loss: 0.0600 - lr: 1.0000e-05 - 268ms/epoch - 6ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8702 - val_loss: 0.0601 - lr: 1.0000e-05 - 246ms/epoch - 6ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8698 - val_loss: 0.0601 - lr: 1.0000e-05 - 248ms/epoch - 6ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8694 - val_loss: 0.0602 - lr: 1.0000e-05 - 249ms/epoch - 6ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00016: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8690 - val_loss: 0.0602 - lr: 1.0000e-05 - 237ms/epoch - 6ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8686 - val_loss: 0.0603 - lr: 1.0000e-05 - 244ms/epoch - 6ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8682 - val_loss: 0.0603 - lr: 1.0000e-05 - 273ms/epoch - 6ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8678 - val_loss: 0.0604 - lr: 1.0000e-05 - 298ms/epoch - 7ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8674 - val_loss: 0.0604 - lr: 1.0000e-05 - 266ms/epoch - 6ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8670 - val_loss: 0.0605 - lr: 1.0000e-05 - 256ms/epoch - 6ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8666 - val_loss: 0.0605 - lr: 1.0000e-05 - 222ms/epoch - 5ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8662 - val_loss: 0.0606 - lr: 1.0000e-05 - 253ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8658 - val_loss: 0.0606 - lr: 1.0000e-05 - 295ms/epoch - 7ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8654 - val_loss: 0.0607 - lr: 1.0000e-05 - 236ms/epoch - 5ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8650 - val_loss: 0.0607 - lr: 1.0000e-05 - 242ms/epoch - 6ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8646 - val_loss: 0.0608 - lr: 1.0000e-05 - 278ms/epoch - 6ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8642 - val_loss: 0.0608 - lr: 1.0000e-05 - 239ms/epoch - 6ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8638 - val_loss: 0.0609 - lr: 1.0000e-05 - 223ms/epoch - 5ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8634 - val_loss: 0.0609 - lr: 1.0000e-05 - 247ms/epoch - 6ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8630 - val_loss: 0.0610 - lr: 1.0000e-05 - 240ms/epoch - 6ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8627 - val_loss: 0.0610 - lr: 1.0000e-05 - 245ms/epoch - 6ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8623 - val_loss: 0.0611 - lr: 1.0000e-05 - 219ms/epoch - 5ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8619 - val_loss: 0.0612 - lr: 1.0000e-05 - 246ms/epoch - 6ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8615 - val_loss: 0.0612 - lr: 1.0000e-05 - 249ms/epoch - 6ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8611 - val_loss: 0.0613 - lr: 1.0000e-05 - 235ms/epoch - 5ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8607 - val_loss: 0.0613 - lr: 1.0000e-05 - 234ms/epoch - 5ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8603 - val_loss: 0.0614 - lr: 1.0000e-05 - 240ms/epoch - 6ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8599 - val_loss: 0.0614 - lr: 1.0000e-05 - 285ms/epoch - 7ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8595 - val_loss: 0.0615 - lr: 1.0000e-05 - 258ms/epoch - 6ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8591 - val_loss: 0.0616 - lr: 1.0000e-05 - 231ms/epoch - 5ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8587 - val_loss: 0.0616 - lr: 1.0000e-05 - 245ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8583 - val_loss: 0.0617 - lr: 1.0000e-05 - 252ms/epoch - 6ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8579 - val_loss: 0.0618 - lr: 1.0000e-05 - 255ms/epoch - 6ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8575 - val_loss: 0.0618 - lr: 1.0000e-05 - 255ms/epoch - 6ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8571 - val_loss: 0.0619 - lr: 1.0000e-05 - 248ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8567 - val_loss: 0.0619 - lr: 1.0000e-05 - 240ms/epoch - 6ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8563 - val_loss: 0.0620 - lr: 1.0000e-05 - 247ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8559 - val_loss: 0.0621 - lr: 1.0000e-05 - 258ms/epoch - 6ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8555 - val_loss: 0.0621 - lr: 1.0000e-05 - 250ms/epoch - 6ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.04608
43/43 - 0s - loss: 0.8552 - val_loss: 0.0622 - lr: 1.0000e-05 - 270ms/epoch - 6ms/step
Epoch 00051: early stopping
SMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 48.51% Accuracy
MSE: 23.82392201920814
RMSE: 4.880975519218278
MAPE: 3.848681389687446
EMA
Prediction vs Close: 54.48% Accuracy
Prediction vs Prediction: 47.76% Accuracy
MSE: 36.51946703896653
RMSE: 6.043133875644865
MAPE: 4.745071953502077
WMA
Prediction vs Close: 53.73% Accuracy
Prediction vs Prediction: 45.52% Accuracy
MSE: 43.8622174139386
RMSE: 6.622855684214975
MAPE: 5.251112001385055
DEMA
Prediction vs Close: 52.99% Accuracy
Prediction vs Prediction: 46.64% Accuracy
MSE: 142.0772798190819
RMSE: 11.91961743593652
MAPE: 10.62322903609799
KAMA
Prediction vs Close: 54.1% Accuracy
Prediction vs Prediction: 50.0% Accuracy
MSE: 22.038946216129663
RMSE: 4.694565604625168
MAPE: 3.7548184287573756
MIDPOINT
Prediction vs Close: 51.87% Accuracy
Prediction vs Prediction: 47.01% Accuracy
MSE: 18.4805779796863
RMSE: 4.298904276636815
MAPE: 3.422385888494848
T3
Prediction vs Close: 53.36% Accuracy
Prediction vs Prediction: 45.9% Accuracy
MSE: 77.67949047095865
RMSE: 8.813596908808496
MAPE: 7.190279750268759
TEMA
TEMA([input_arrays], [timeperiod=30])
Triple Exponential Moving Average (Overlap Studies)
Inputs:
price: (any ndarray)
Parameters:
timeperiod: 30
Outputs:
real
9
Working on TEMA predictions
parameters used : 808 269
Performing stepwise search to minimize aic
ARIMA(1,3,1)(0,0,0)[0] : AIC=-16762.799, Time=5.08 sec
ARIMA(0,3,0)(0,0,0)[0] : AIC=-14158.507, Time=2.96 sec
ARIMA(1,3,0)(0,0,0)[0] : AIC=-16445.598, Time=9.14 sec
ARIMA(0,3,1)(0,0,0)[0] : AIC=-16144.282, Time=11.83 sec
ARIMA(2,3,1)(0,0,0)[0] : AIC=-15275.101, Time=9.02 sec
ARIMA(1,3,2)(0,0,0)[0] : AIC=-15897.090, Time=13.05 sec
ARIMA(0,3,2)(0,0,0)[0] : AIC=-16446.973, Time=9.39 sec
ARIMA(2,3,0)(0,0,0)[0] : AIC=-16567.628, Time=3.59 sec
/usr/local/lib/python3.7/dist-packages/statsmodels/tsa/statespace/sarimax.py:1906: RuntimeWarning: divide by zero encountered in reciprocal return np.roots(self.polynomial_reduced_ma)**-1
ARIMA(2,3,2)(0,0,0)[0] : AIC=-16523.926, Time=3.73 sec
ARIMA(1,3,1)(0,0,0)[0] intercept : AIC=-16696.008, Time=3.66 sec
Best model: ARIMA(1,3,1)(0,0,0)[0]
Total fit time: 71.466 seconds
SARIMAX Results
==============================================================================
Dep. Variable: y No. Observations: 808
Model: SARIMAX(1, 3, 1) Log Likelihood 8408.400
Date: Sun, 12 Dec 2021 AIC -16762.799
Time: 18:47:46 BIC -16636.147
Sample: 0 HQIC -16714.159
- 808
Covariance Type: opg
==============================================================================
coef std err z P>|z| [0.025 0.975]
------------------------------------------------------------------------------
x1 -5.289e-07 0.001 -0.000 1.000 -0.002 0.002
x2 -5.288e-07 0.001 -0.001 0.999 -0.002 0.002
x3 -5.306e-07 0.001 -0.000 1.000 -0.002 0.002
x4 1.0000 0.000 2045.695 0.000 0.999 1.001
x5 -5.041e-07 0.000 -0.001 0.999 -0.001 0.001
x6 -9.879e-07 4.33e-05 -0.023 0.982 -8.58e-05 8.38e-05
x7 -5.185e-07 0.001 -0.001 0.999 -0.001 0.001
x8 0.0001 0.000 0.643 0.520 -0.000 0.001
x9 9.794e-08 0.001 0.000 1.000 -0.001 0.001
x10 0.0001 0.000 0.313 0.754 -0.001 0.001
x11 -0.0004 0.000 -2.284 0.022 -0.001 -6.06e-05
x12 0.0005 0.000 2.453 0.014 0.000 0.001
x13 -5.277e-07 0.000 -0.002 0.999 -0.001 0.001
x14 -1.566e-06 0.000 -0.005 0.996 -0.001 0.001
x15 -5.136e-07 9.86e-05 -0.005 0.996 -0.000 0.000
x16 -7.66e-07 0.000 -0.002 0.999 -0.001 0.001
x17 -5.146e-07 0.000 -0.003 0.998 -0.000 0.000
x18 -1.701e-07 0.001 -0.000 1.000 -0.001 0.001
x19 -5.77e-07 8.54e-05 -0.007 0.995 -0.000 0.000
x20 5.026e-07 0.001 0.001 0.999 -0.001 0.001
x21 -2.058e-06 0.000 -0.010 0.992 -0.000 0.000
x22 -1.098e-06 0.001 -0.001 0.999 -0.003 0.003
x23 -1.472e-06 0.001 -0.003 0.998 -0.001 0.001
x24 -8.255e-07 0.001 -0.001 0.999 -0.002 0.002
ar.L1 -0.2866 3.63e-05 -7897.273 0.000 -0.287 -0.287
ma.L1 -0.9124 1.46e-06 -6.25e+05 0.000 -0.912 -0.912
sigma2 9.98e-11 7.23e-11 1.380 0.168 -4.2e-11 2.42e-10
===================================================================================
Ljung-Box (L1) (Q): 83.51 Jarque-Bera (JB): 4742889.91
Prob(Q): 0.00 Prob(JB): 0.00
Heteroskedasticity (H): 0.00 Skew: -5.71
Prob(H) (two-sided): 0.00 Kurtosis: 378.86
===================================================================================
Warnings:
[1] Covariance matrix calculated using the outer product of gradients (complex-step).
[2] Covariance matrix is singular or near-singular, with condition number 4.2e+22. Standard errors may be unstable.
ARIMA order: (1, 3, 1)
Epoch 1/500
Epoch 00001: val_loss improved from inf to 0.04686, saving model to LSTM8.h5
90/90 - 4s - loss: 1.2088 - val_loss: 0.0469 - lr: 0.0010 - 4s/epoch - 45ms/step
Epoch 2/500
Epoch 00002: val_loss did not improve from 0.04686
90/90 - 0s - loss: 1.0094 - val_loss: 0.0553 - lr: 0.0010 - 456ms/epoch - 5ms/step
Epoch 3/500
Epoch 00003: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.8756 - val_loss: 0.0627 - lr: 0.0010 - 484ms/epoch - 5ms/step
Epoch 4/500
Epoch 00004: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.7962 - val_loss: 0.0702 - lr: 0.0010 - 477ms/epoch - 5ms/step
Epoch 5/500
Epoch 00005: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.7458 - val_loss: 0.0777 - lr: 0.0010 - 543ms/epoch - 6ms/step
Epoch 6/500
Epoch 00006: ReduceLROnPlateau reducing learning rate to 0.00010000000474974513.
Epoch 00006: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.7107 - val_loss: 0.0852 - lr: 0.0010 - 549ms/epoch - 6ms/step
Epoch 7/500
Epoch 00007: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6938 - val_loss: 0.0859 - lr: 1.0000e-04 - 477ms/epoch - 5ms/step
Epoch 8/500
Epoch 00008: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6912 - val_loss: 0.0867 - lr: 1.0000e-04 - 532ms/epoch - 6ms/step
Epoch 9/500
Epoch 00009: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6886 - val_loss: 0.0876 - lr: 1.0000e-04 - 472ms/epoch - 5ms/step
Epoch 10/500
Epoch 00010: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6860 - val_loss: 0.0884 - lr: 1.0000e-04 - 564ms/epoch - 6ms/step
Epoch 11/500
Epoch 00011: ReduceLROnPlateau reducing learning rate to 1.0000000474974514e-05.
Epoch 00011: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6834 - val_loss: 0.0893 - lr: 1.0000e-04 - 473ms/epoch - 5ms/step
Epoch 12/500
Epoch 00012: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6818 - val_loss: 0.0894 - lr: 1.0000e-05 - 518ms/epoch - 6ms/step
Epoch 13/500
Epoch 00013: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6815 - val_loss: 0.0895 - lr: 1.0000e-05 - 554ms/epoch - 6ms/step
Epoch 14/500
Epoch 00014: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6812 - val_loss: 0.0896 - lr: 1.0000e-05 - 458ms/epoch - 5ms/step
Epoch 15/500
Epoch 00015: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6810 - val_loss: 0.0897 - lr: 1.0000e-05 - 443ms/epoch - 5ms/step
Epoch 16/500
Epoch 00016: ReduceLROnPlateau reducing learning rate to 1e-05.
Epoch 00016: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6807 - val_loss: 0.0899 - lr: 1.0000e-05 - 451ms/epoch - 5ms/step
Epoch 17/500
Epoch 00017: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6804 - val_loss: 0.0900 - lr: 1.0000e-05 - 469ms/epoch - 5ms/step
Epoch 18/500
Epoch 00018: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6801 - val_loss: 0.0901 - lr: 1.0000e-05 - 451ms/epoch - 5ms/step
Epoch 19/500
Epoch 00019: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6798 - val_loss: 0.0902 - lr: 1.0000e-05 - 452ms/epoch - 5ms/step
Epoch 20/500
Epoch 00020: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6795 - val_loss: 0.0903 - lr: 1.0000e-05 - 431ms/epoch - 5ms/step
Epoch 21/500
Epoch 00021: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6791 - val_loss: 0.0905 - lr: 1.0000e-05 - 548ms/epoch - 6ms/step
Epoch 22/500
Epoch 00022: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6788 - val_loss: 0.0906 - lr: 1.0000e-05 - 443ms/epoch - 5ms/step
Epoch 23/500
Epoch 00023: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6785 - val_loss: 0.0907 - lr: 1.0000e-05 - 539ms/epoch - 6ms/step
Epoch 24/500
Epoch 00024: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6782 - val_loss: 0.0909 - lr: 1.0000e-05 - 528ms/epoch - 6ms/step
Epoch 25/500
Epoch 00025: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6778 - val_loss: 0.0910 - lr: 1.0000e-05 - 453ms/epoch - 5ms/step
Epoch 26/500
Epoch 00026: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6775 - val_loss: 0.0911 - lr: 1.0000e-05 - 506ms/epoch - 6ms/step
Epoch 27/500
Epoch 00027: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6772 - val_loss: 0.0913 - lr: 1.0000e-05 - 455ms/epoch - 5ms/step
Epoch 28/500
Epoch 00028: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6768 - val_loss: 0.0914 - lr: 1.0000e-05 - 514ms/epoch - 6ms/step
Epoch 29/500
Epoch 00029: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6765 - val_loss: 0.0916 - lr: 1.0000e-05 - 478ms/epoch - 5ms/step
Epoch 30/500
Epoch 00030: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6762 - val_loss: 0.0917 - lr: 1.0000e-05 - 514ms/epoch - 6ms/step
Epoch 31/500
Epoch 00031: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6758 - val_loss: 0.0919 - lr: 1.0000e-05 - 535ms/epoch - 6ms/step
Epoch 32/500
Epoch 00032: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6755 - val_loss: 0.0921 - lr: 1.0000e-05 - 432ms/epoch - 5ms/step
Epoch 33/500
Epoch 00033: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6751 - val_loss: 0.0922 - lr: 1.0000e-05 - 543ms/epoch - 6ms/step
Epoch 34/500
Epoch 00034: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6748 - val_loss: 0.0924 - lr: 1.0000e-05 - 454ms/epoch - 5ms/step
Epoch 35/500
Epoch 00035: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6744 - val_loss: 0.0925 - lr: 1.0000e-05 - 481ms/epoch - 5ms/step
Epoch 36/500
Epoch 00036: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6741 - val_loss: 0.0927 - lr: 1.0000e-05 - 533ms/epoch - 6ms/step
Epoch 37/500
Epoch 00037: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6738 - val_loss: 0.0929 - lr: 1.0000e-05 - 574ms/epoch - 6ms/step
Epoch 38/500
Epoch 00038: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6734 - val_loss: 0.0930 - lr: 1.0000e-05 - 498ms/epoch - 6ms/step
Epoch 39/500
Epoch 00039: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6731 - val_loss: 0.0932 - lr: 1.0000e-05 - 486ms/epoch - 5ms/step
Epoch 40/500
Epoch 00040: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6727 - val_loss: 0.0934 - lr: 1.0000e-05 - 467ms/epoch - 5ms/step
Epoch 41/500
Epoch 00041: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6724 - val_loss: 0.0936 - lr: 1.0000e-05 - 540ms/epoch - 6ms/step
Epoch 42/500
Epoch 00042: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6720 - val_loss: 0.0937 - lr: 1.0000e-05 - 577ms/epoch - 6ms/step
Epoch 43/500
Epoch 00043: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6717 - val_loss: 0.0939 - lr: 1.0000e-05 - 461ms/epoch - 5ms/step
Epoch 44/500
Epoch 00044: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6713 - val_loss: 0.0941 - lr: 1.0000e-05 - 458ms/epoch - 5ms/step
Epoch 45/500
Epoch 00045: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6710 - val_loss: 0.0943 - lr: 1.0000e-05 - 541ms/epoch - 6ms/step
Epoch 46/500
Epoch 00046: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6706 - val_loss: 0.0945 - lr: 1.0000e-05 - 549ms/epoch - 6ms/step
Epoch 47/500
Epoch 00047: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6703 - val_loss: 0.0947 - lr: 1.0000e-05 - 512ms/epoch - 6ms/step
Epoch 48/500
Epoch 00048: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6699 - val_loss: 0.0948 - lr: 1.0000e-05 - 525ms/epoch - 6ms/step
Epoch 49/500
Epoch 00049: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6696 - val_loss: 0.0950 - lr: 1.0000e-05 - 460ms/epoch - 5ms/step
Epoch 50/500
Epoch 00050: val_loss did not improve from 0.04686
90/90 - 1s - loss: 0.6692 - val_loss: 0.0952 - lr: 1.0000e-05 - 558ms/epoch - 6ms/step
Epoch 51/500
Epoch 00051: val_loss did not improve from 0.04686
90/90 - 0s - loss: 0.6689 - val_loss: 0.0954 - lr: 1.0000e-05 - 450ms/epoch - 5ms/step
Epoch 00051: early stopping
SMA Prediction vs Close: 54.1% Accuracy Prediction vs Prediction: 48.51% Accuracy MSE: 23.82392201920814 RMSE: 4.880975519218278 MAPE: 3.848681389687446 EMA Prediction vs Close: 54.48% Accuracy Prediction vs Prediction: 47.76% Accuracy MSE: 36.51946703896653 RMSE: 6.043133875644865 MAPE: 4.745071953502077 WMA Prediction vs Close: 53.73% Accuracy Prediction vs Prediction: 45.52% Accuracy MSE: 43.8622174139386 RMSE: 6.622855684214975 MAPE: 5.251112001385055 DEMA Prediction vs Close: 52.99% Accuracy Prediction vs Prediction: 46.64% Accuracy MSE: 142.0772798190819 RMSE: 11.91961743593652 MAPE: 10.62322903609799 KAMA Prediction vs Close: 54.1% Accuracy Prediction vs Prediction: 50.0% Accuracy MSE: 22.038946216129663 RMSE: 4.694565604625168 MAPE: 3.7548184287573756 MIDPOINT Prediction vs Close: 51.87% Accuracy Prediction vs Prediction: 47.01% Accuracy MSE: 18.4805779796863 RMSE: 4.298904276636815 MAPE: 3.422385888494848 T3 Prediction vs Close: 53.36% Accuracy Prediction vs Prediction: 45.9% Accuracy MSE: 77.67949047095865 RMSE: 8.813596908808496 MAPE: 7.190279750268759 TEMA Prediction vs Close: 51.87% Accuracy Prediction vs Prediction: 48.13% Accuracy MSE: 31.990355699152573 RMSE: 5.656001741438255 MAPE: 5.124316314221195 Runtime: mins: 54.23975520736664
Architecture Used¶
In [140]:
from google.colab import files
import cv2
uploaded = files.upload()
Saving Experiment8.png to Experiment8 (1).png
In [143]:
img = cv2.imread('Experiment8.png')
plt.figure(figsize=(20,10))
plt.axis("off")
plt.title('LSTM Architecture '+imgfile,fontsize=18)
plt.imshow(img)
Out[143]:
<matplotlib.image.AxesImage at 0x7f75c4377150>
Model Plots¶
In [112]:
with open('simulation8_data.json') as json_file:
simulation8 = json.load(json_file)
fileimg = 'Experiment8'
In [113]:
for i in range(len(list(simulation8.keys()))):
SIM = list(simulation8.keys())[i]
plot_train(simulation8,SIM)
plot_test(simulation8,SIM)
----- Train RMSE for SMA ----- 18.846065584784647 ----- Train_MSE_LSTM for SMA ----- 355.1741880260043 ----- Train MAE LSTM for SMA ----- 18.820755524210412
----- Test RMSE for SMA----- 4.880975519218278 ----- Test_MSE_LSTM for SMA----- 23.82392201920814 ----- Test_MAE_LSTM for SMA----- 3.848681389687446
----- Train RMSE for EMA ----- 24.364485496157467 ----- Train_MSE_LSTM for EMA ----- 593.6281534924675 ----- Train MAE LSTM for EMA ----- 24.35965843720011
----- Test RMSE for EMA----- 6.043133875644865 ----- Test_MSE_LSTM for EMA----- 36.51946703896653 ----- Test_MAE_LSTM for EMA----- 4.745071953502077
----- Train RMSE for WMA ----- 22.522819506061108 ----- Train_MSE_LSTM for WMA ----- 507.27739850260673 ----- Train MAE LSTM for WMA ----- 22.511068136385173
----- Test RMSE for WMA----- 6.622855684214975 ----- Test_MSE_LSTM for WMA----- 43.8622174139386 ----- Test_MAE_LSTM for WMA----- 5.251112001385055
----- Train RMSE for DEMA ----- 28.644253182641428 ----- Train_MSE_LSTM for DEMA ----- 820.4932403912636 ----- Train MAE LSTM for DEMA ----- 28.644042274739483
----- Test RMSE for DEMA----- 11.91961743593652 ----- Test_MSE_LSTM for DEMA----- 142.0772798190819 ----- Test_MAE_LSTM for DEMA----- 10.62322903609799
----- Train RMSE for KAMA ----- 18.988700576595924 ----- Train_MSE_LSTM for KAMA ----- 360.5707495876144 ----- Train MAE LSTM for KAMA ----- 18.970468660392385
----- Test RMSE for KAMA----- 4.694565604625168 ----- Test_MSE_LSTM for KAMA----- 22.038946216129663 ----- Test_MAE_LSTM for KAMA----- 3.7548184287573756
----- Train RMSE for MIDPOINT ----- 16.06235031499391 ----- Train_MSE_LSTM for MIDPOINT ----- 257.999097641585 ----- Train MAE LSTM for MIDPOINT ----- 15.981051393074564
----- Test RMSE for MIDPOINT----- 4.298904276636815 ----- Test_MSE_LSTM for MIDPOINT----- 18.4805779796863 ----- Test_MAE_LSTM for MIDPOINT----- 3.422385888494848
----- Train RMSE for T3 ----- 20.284725351435558 ----- Train_MSE_LSTM for T3 ----- 411.47008258317237 ----- Train MAE LSTM for T3 ----- 20.220124100694562
----- Test RMSE for T3----- 8.813596908808496 ----- Test_MSE_LSTM for T3----- 77.67949047095865 ----- Test_MAE_LSTM for T3----- 7.190279750268759
----- Train RMSE for TEMA ----- 18.95244476174229 ----- Train_MSE_LSTM for TEMA ----- 359.19516244689277 ----- Train MAE LSTM for TEMA ----- 18.91601826176785
----- Test RMSE for TEMA----- 5.656001741438255 ----- Test_MSE_LSTM for TEMA----- 31.990355699152573 ----- Test_MAE_LSTM for TEMA----- 5.124316314221195
List of RMSE, MSE & MAE scores for Test data¶
In [5]:
import json
with open('simulation1_data.json') as json_file:
simulation1 = json.load(json_file)
with open('simulation2_data.json') as json_file:
simulation2 = json.load(json_file)
with open('simulation3_data.json') as json_file:
simulation3 = json.load(json_file)
with open('simulation4_data.json') as json_file:
simulation4 = json.load(json_file)
with open('simulation5_data.json') as json_file:
simulation5 = json.load(json_file)
with open('simulation6_data.json') as json_file:
simulation6 = json.load(json_file)
with open('simulation7_data.json') as json_file:
simulation7 = json.load(json_file)
with open('simulation8_data.json') as json_file:
simulation8 = json.load(json_file)
In [6]:
text = 'Stock with Sentiment '
simulations = [simulation1,simulation2,simulation3,simulation4,simulation5,simulation6,simulation7,simulation8]
for i,simulation in enumerate(simulations):
for ma in simulation.keys():
print(text+'Experiment ',i+1,' for MA :',ma,'the MSE is: ',simulation[ma]['final']['mse'])
print(text+'Experiment ',i+1,' for MA :',ma,'the RMSE is: ',simulation[ma]['final']['rmse'])
print(text+'Experiment ',i+1,' for MA :',ma,'the MAE is: ',simulation[ma]['final']['mae'])
Stock with Sentiment Experiment 1 for MA : SMA the MSE is: 39.29635655476263 Stock with Sentiment Experiment 1 for MA : SMA the RMSE is: 6.268680607174258 Stock with Sentiment Experiment 1 for MA : SMA the MAE is: 5.07416987111494 Stock with Sentiment Experiment 1 for MA : EMA the MSE is: 30.25057164247689 Stock with Sentiment Experiment 1 for MA : EMA the RMSE is: 5.50005196725239 Stock with Sentiment Experiment 1 for MA : EMA the MAE is: 4.444486270049439 Stock with Sentiment Experiment 1 for MA : WMA the MSE is: 64.92643731055051 Stock with Sentiment Experiment 1 for MA : WMA the RMSE is: 8.057694292448089 Stock with Sentiment Experiment 1 for MA : WMA the MAE is: 6.289003841800032 Stock with Sentiment Experiment 1 for MA : DEMA the MSE is: 52.98651621196421 Stock with Sentiment Experiment 1 for MA : DEMA the RMSE is: 7.279183760008 Stock with Sentiment Experiment 1 for MA : DEMA the MAE is: 5.725540843661134 Stock with Sentiment Experiment 1 for MA : KAMA the MSE is: 33.6026987861483 Stock with Sentiment Experiment 1 for MA : KAMA the RMSE is: 5.796783486223054 Stock with Sentiment Experiment 1 for MA : KAMA the MAE is: 4.518981487962124 Stock with Sentiment Experiment 1 for MA : MIDPOINT the MSE is: 22.7414116550439 Stock with Sentiment Experiment 1 for MA : MIDPOINT the RMSE is: 4.768795618921396 Stock with Sentiment Experiment 1 for MA : MIDPOINT the MAE is: 3.944458615157319 Stock with Sentiment Experiment 1 for MA : T3 the MSE is: 71.14601272964887 Stock with Sentiment Experiment 1 for MA : T3 the RMSE is: 8.434809584670473 Stock with Sentiment Experiment 1 for MA : T3 the MAE is: 6.848574357624394 Stock with Sentiment Experiment 1 for MA : TEMA the MSE is: 26.608302345887367 Stock with Sentiment Experiment 1 for MA : TEMA the RMSE is: 5.158323598407468 Stock with Sentiment Experiment 1 for MA : TEMA the MAE is: 4.336839144940602 Stock with Sentiment Experiment 2 for MA : SMA the MSE is: 39.87524298067742 Stock with Sentiment Experiment 2 for MA : SMA the RMSE is: 6.3146847095225125 Stock with Sentiment Experiment 2 for MA : SMA the MAE is: 5.088204561858354 Stock with Sentiment Experiment 2 for MA : EMA the MSE is: 73.6938994303845 Stock with Sentiment Experiment 2 for MA : EMA the RMSE is: 8.584515095821342 Stock with Sentiment Experiment 2 for MA : EMA the MAE is: 7.207683534137507 Stock with Sentiment Experiment 2 for MA : WMA the MSE is: 76.90233333404232 Stock with Sentiment Experiment 2 for MA : WMA the RMSE is: 8.76939754681257 Stock with Sentiment Experiment 2 for MA : WMA the MAE is: 7.121360770950225 Stock with Sentiment Experiment 2 for MA : DEMA the MSE is: 131.0602137141292 Stock with Sentiment Experiment 2 for MA : DEMA the RMSE is: 11.448153288374908 Stock with Sentiment Experiment 2 for MA : DEMA the MAE is: 10.329401784343453 Stock with Sentiment Experiment 2 for MA : KAMA the MSE is: 42.128331463904196 Stock with Sentiment Experiment 2 for MA : KAMA the RMSE is: 6.49063413418937 Stock with Sentiment Experiment 2 for MA : KAMA the MAE is: 5.225311201416733 Stock with Sentiment Experiment 2 for MA : MIDPOINT the MSE is: 102.72361049298229 Stock with Sentiment Experiment 2 for MA : MIDPOINT the RMSE is: 10.135265684380567 Stock with Sentiment Experiment 2 for MA : MIDPOINT the MAE is: 8.372939983805384 Stock with Sentiment Experiment 2 for MA : T3 the MSE is: 106.78678832331099 Stock with Sentiment Experiment 2 for MA : T3 the RMSE is: 10.333769318274479 Stock with Sentiment Experiment 2 for MA : T3 the MAE is: 8.334873887974807 Stock with Sentiment Experiment 2 for MA : TEMA the MSE is: 78.66857412996737 Stock with Sentiment Experiment 2 for MA : TEMA the RMSE is: 8.86953066007257 Stock with Sentiment Experiment 2 for MA : TEMA the MAE is: 7.566655356961055 Stock with Sentiment Experiment 3 for MA : SMA the MSE is: 39.71707698339657 Stock with Sentiment Experiment 3 for MA : SMA the RMSE is: 6.302148600548591 Stock with Sentiment Experiment 3 for MA : SMA the MAE is: 5.205453851116191 Stock with Sentiment Experiment 3 for MA : EMA the MSE is: 18.546398408821787 Stock with Sentiment Experiment 3 for MA : EMA the RMSE is: 4.306552961339474 Stock with Sentiment Experiment 3 for MA : EMA the MAE is: 3.4160340524918316 Stock with Sentiment Experiment 3 for MA : WMA the MSE is: 100.42411916020163 Stock with Sentiment Experiment 3 for MA : WMA the RMSE is: 10.021183520932126 Stock with Sentiment Experiment 3 for MA : WMA the MAE is: 8.070563561893302 Stock with Sentiment Experiment 3 for MA : DEMA the MSE is: 42.724210554814 Stock with Sentiment Experiment 3 for MA : DEMA the RMSE is: 6.536375949623308 Stock with Sentiment Experiment 3 for MA : DEMA the MAE is: 5.453197818871469 Stock with Sentiment Experiment 3 for MA : KAMA the MSE is: 23.65848136204503 Stock with Sentiment Experiment 3 for MA : KAMA the RMSE is: 4.863998495275778 Stock with Sentiment Experiment 3 for MA : KAMA the MAE is: 3.972687129543795 Stock with Sentiment Experiment 3 for MA : MIDPOINT the MSE is: 21.13528196098577 Stock with Sentiment Experiment 3 for MA : MIDPOINT the RMSE is: 4.5973124715409295 Stock with Sentiment Experiment 3 for MA : MIDPOINT the MAE is: 3.8079919461917573 Stock with Sentiment Experiment 3 for MA : T3 the MSE is: 101.89511494721017 Stock with Sentiment Experiment 3 for MA : T3 the RMSE is: 10.09431101894578 Stock with Sentiment Experiment 3 for MA : T3 the MAE is: 8.111172771475383 Stock with Sentiment Experiment 3 for MA : TEMA the MSE is: 10.999022660597852 Stock with Sentiment Experiment 3 for MA : TEMA the RMSE is: 3.316477447623887 Stock with Sentiment Experiment 3 for MA : TEMA the MAE is: 2.655228978622383 Stock with Sentiment Experiment 4 for MA : SMA the MSE is: 25.081351909450348 Stock with Sentiment Experiment 4 for MA : SMA the RMSE is: 5.008128583557969 Stock with Sentiment Experiment 4 for MA : SMA the MAE is: 3.9377037384058786 Stock with Sentiment Experiment 4 for MA : EMA the MSE is: 37.86506779592453 Stock with Sentiment Experiment 4 for MA : EMA the RMSE is: 6.153459823215273 Stock with Sentiment Experiment 4 for MA : EMA the MAE is: 4.830217084369749 Stock with Sentiment Experiment 4 for MA : WMA the MSE is: 53.37812690087662 Stock with Sentiment Experiment 4 for MA : WMA the RMSE is: 7.306033595657539 Stock with Sentiment Experiment 4 for MA : WMA the MAE is: 5.95316326588041 Stock with Sentiment Experiment 4 for MA : DEMA the MSE is: 133.19678136227444 Stock with Sentiment Experiment 4 for MA : DEMA the RMSE is: 11.541090995320783 Stock with Sentiment Experiment 4 for MA : DEMA the MAE is: 10.29859546777107 Stock with Sentiment Experiment 4 for MA : KAMA the MSE is: 20.693935177088164 Stock with Sentiment Experiment 4 for MA : KAMA the RMSE is: 4.549058713304123 Stock with Sentiment Experiment 4 for MA : KAMA the MAE is: 3.6577262429810227 Stock with Sentiment Experiment 4 for MA : MIDPOINT the MSE is: 18.24421544500263 Stock with Sentiment Experiment 4 for MA : MIDPOINT the RMSE is: 4.271324788049093 Stock with Sentiment Experiment 4 for MA : MIDPOINT the MAE is: 3.3887721441386436 Stock with Sentiment Experiment 4 for MA : T3 the MSE is: 74.9216743993189 Stock with Sentiment Experiment 4 for MA : T3 the RMSE is: 8.655730725901707 Stock with Sentiment Experiment 4 for MA : T3 the MAE is: 7.03412901576443 Stock with Sentiment Experiment 4 for MA : TEMA the MSE is: 33.63623508699449 Stock with Sentiment Experiment 4 for MA : TEMA the RMSE is: 5.799675429452453 Stock with Sentiment Experiment 4 for MA : TEMA the MAE is: 5.291282498785388 Stock with Sentiment Experiment 5 for MA : SMA the MSE is: 22.905457367129987 Stock with Sentiment Experiment 5 for MA : SMA the RMSE is: 4.785964622427749 Stock with Sentiment Experiment 5 for MA : SMA the MAE is: 4.003866138329542 Stock with Sentiment Experiment 5 for MA : EMA the MSE is: 24.329194491765858 Stock with Sentiment Experiment 5 for MA : EMA the RMSE is: 4.932463328983385 Stock with Sentiment Experiment 5 for MA : EMA the MAE is: 4.099346553838579 Stock with Sentiment Experiment 5 for MA : WMA the MSE is: 26.731339639305343 Stock with Sentiment Experiment 5 for MA : WMA the RMSE is: 5.170235936522176 Stock with Sentiment Experiment 5 for MA : WMA the MAE is: 4.142288801040536 Stock with Sentiment Experiment 5 for MA : DEMA the MSE is: 20.555600166870196 Stock with Sentiment Experiment 5 for MA : DEMA the RMSE is: 4.53382842274277 Stock with Sentiment Experiment 5 for MA : DEMA the MAE is: 3.6522177332314283 Stock with Sentiment Experiment 5 for MA : KAMA the MSE is: 36.658985139129086 Stock with Sentiment Experiment 5 for MA : KAMA the RMSE is: 6.054666393710646 Stock with Sentiment Experiment 5 for MA : KAMA the MAE is: 4.91375972579294 Stock with Sentiment Experiment 5 for MA : MIDPOINT the MSE is: 46.211728483578064 Stock with Sentiment Experiment 5 for MA : MIDPOINT the RMSE is: 6.797920894183608 Stock with Sentiment Experiment 5 for MA : MIDPOINT the MAE is: 5.510818514624332 Stock with Sentiment Experiment 5 for MA : T3 the MSE is: 39.422414094648474 Stock with Sentiment Experiment 5 for MA : T3 the RMSE is: 6.278727107833918 Stock with Sentiment Experiment 5 for MA : T3 the MAE is: 5.177910469752962 Stock with Sentiment Experiment 5 for MA : TEMA the MSE is: 27.219221961342864 Stock with Sentiment Experiment 5 for MA : TEMA the RMSE is: 5.217204420122223 Stock with Sentiment Experiment 5 for MA : TEMA the MAE is: 4.028826161355838 Stock with Sentiment Experiment 6 for MA : SMA the MSE is: 109.05489356902385 Stock with Sentiment Experiment 6 for MA : SMA the RMSE is: 10.442935103170173 Stock with Sentiment Experiment 6 for MA : SMA the MAE is: 8.732625329283675 Stock with Sentiment Experiment 6 for MA : EMA the MSE is: 73.04142380966745 Stock with Sentiment Experiment 6 for MA : EMA the RMSE is: 8.546427546622475 Stock with Sentiment Experiment 6 for MA : EMA the MAE is: 7.099244401385842 Stock with Sentiment Experiment 6 for MA : WMA the MSE is: 84.94978866341171 Stock with Sentiment Experiment 6 for MA : WMA the RMSE is: 9.21682096296829 Stock with Sentiment Experiment 6 for MA : WMA the MAE is: 7.490547440692417 Stock with Sentiment Experiment 6 for MA : DEMA the MSE is: 151.61070955572364 Stock with Sentiment Experiment 6 for MA : DEMA the RMSE is: 12.313030072070955 Stock with Sentiment Experiment 6 for MA : DEMA the MAE is: 11.085595013418024 Stock with Sentiment Experiment 6 for MA : KAMA the MSE is: 80.0014001509936 Stock with Sentiment Experiment 6 for MA : KAMA the RMSE is: 8.944350180476702 Stock with Sentiment Experiment 6 for MA : KAMA the MAE is: 7.358601729961256 Stock with Sentiment Experiment 6 for MA : MIDPOINT the MSE is: 61.18379984959283 Stock with Sentiment Experiment 6 for MA : MIDPOINT the RMSE is: 7.822007405365507 Stock with Sentiment Experiment 6 for MA : MIDPOINT the MAE is: 6.441728946960992 Stock with Sentiment Experiment 6 for MA : T3 the MSE is: 110.56518298054853 Stock with Sentiment Experiment 6 for MA : T3 the RMSE is: 10.514998001927937 Stock with Sentiment Experiment 6 for MA : T3 the MAE is: 8.473394546481362 Stock with Sentiment Experiment 6 for MA : TEMA the MSE is: 69.56753550271695 Stock with Sentiment Experiment 6 for MA : TEMA the RMSE is: 8.340715527022663 Stock with Sentiment Experiment 6 for MA : TEMA the MAE is: 7.185876850367952 Stock with Sentiment Experiment 7 for MA : SMA the MSE is: 133.6684179910327 Stock with Sentiment Experiment 7 for MA : SMA the RMSE is: 11.561505870388714 Stock with Sentiment Experiment 7 for MA : SMA the MAE is: 10.289389775089397 Stock with Sentiment Experiment 7 for MA : EMA the MSE is: 34.52931938659473 Stock with Sentiment Experiment 7 for MA : EMA the RMSE is: 5.876165364129462 Stock with Sentiment Experiment 7 for MA : EMA the MAE is: 4.852473639818076 Stock with Sentiment Experiment 7 for MA : WMA the MSE is: 36.74468487644727 Stock with Sentiment Experiment 7 for MA : WMA the RMSE is: 6.061739426637149 Stock with Sentiment Experiment 7 for MA : WMA the MAE is: 4.85767480758183 Stock with Sentiment Experiment 7 for MA : DEMA the MSE is: 87.45802496937279 Stock with Sentiment Experiment 7 for MA : DEMA the RMSE is: 9.351899538028238 Stock with Sentiment Experiment 7 for MA : DEMA the MAE is: 8.239361009856534 Stock with Sentiment Experiment 7 for MA : KAMA the MSE is: 55.655998887145394 Stock with Sentiment Experiment 7 for MA : KAMA the RMSE is: 7.460294825752223 Stock with Sentiment Experiment 7 for MA : KAMA the MAE is: 6.325008398714769 Stock with Sentiment Experiment 7 for MA : MIDPOINT the MSE is: 28.58496814832563 Stock with Sentiment Experiment 7 for MA : MIDPOINT the RMSE is: 5.346491199686541 Stock with Sentiment Experiment 7 for MA : MIDPOINT the MAE is: 4.412567377496203 Stock with Sentiment Experiment 7 for MA : T3 the MSE is: 127.72960460138046 Stock with Sentiment Experiment 7 for MA : T3 the RMSE is: 11.301752280128088 Stock with Sentiment Experiment 7 for MA : T3 the MAE is: 9.16046603172084 Stock with Sentiment Experiment 7 for MA : TEMA the MSE is: 34.19976389234231 Stock with Sentiment Experiment 7 for MA : TEMA the RMSE is: 5.8480564200717415 Stock with Sentiment Experiment 7 for MA : TEMA the MAE is: 5.017065986818192 Stock with Sentiment Experiment 8 for MA : SMA the MSE is: 23.82392201920814 Stock with Sentiment Experiment 8 for MA : SMA the RMSE is: 4.880975519218278 Stock with Sentiment Experiment 8 for MA : SMA the MAE is: 3.848681389687446 Stock with Sentiment Experiment 8 for MA : EMA the MSE is: 36.51946703896653 Stock with Sentiment Experiment 8 for MA : EMA the RMSE is: 6.043133875644865 Stock with Sentiment Experiment 8 for MA : EMA the MAE is: 4.745071953502077 Stock with Sentiment Experiment 8 for MA : WMA the MSE is: 43.8622174139386 Stock with Sentiment Experiment 8 for MA : WMA the RMSE is: 6.622855684214975 Stock with Sentiment Experiment 8 for MA : WMA the MAE is: 5.251112001385055 Stock with Sentiment Experiment 8 for MA : DEMA the MSE is: 142.0772798190819 Stock with Sentiment Experiment 8 for MA : DEMA the RMSE is: 11.91961743593652 Stock with Sentiment Experiment 8 for MA : DEMA the MAE is: 10.62322903609799 Stock with Sentiment Experiment 8 for MA : KAMA the MSE is: 22.038946216129663 Stock with Sentiment Experiment 8 for MA : KAMA the RMSE is: 4.694565604625168 Stock with Sentiment Experiment 8 for MA : KAMA the MAE is: 3.7548184287573756 Stock with Sentiment Experiment 8 for MA : MIDPOINT the MSE is: 18.4805779796863 Stock with Sentiment Experiment 8 for MA : MIDPOINT the RMSE is: 4.298904276636815 Stock with Sentiment Experiment 8 for MA : MIDPOINT the MAE is: 3.422385888494848 Stock with Sentiment Experiment 8 for MA : T3 the MSE is: 77.67949047095865 Stock with Sentiment Experiment 8 for MA : T3 the RMSE is: 8.813596908808496 Stock with Sentiment Experiment 8 for MA : T3 the MAE is: 7.190279750268759 Stock with Sentiment Experiment 8 for MA : TEMA the MSE is: 31.990355699152573 Stock with Sentiment Experiment 8 for MA : TEMA the RMSE is: 5.656001741438255 Stock with Sentiment Experiment 8 for MA : TEMA the MAE is: 5.124316314221195
In [7]:
text = 'Stock with Sentiment Trends '
simulations = [simulation1,simulation2,simulation3,simulation4,simulation5,simulation6,simulation7,simulation8]
for i,simulation in enumerate(simulations):
for ma in simulation.keys():
# print(text+'Experiment ',i+1,' for MA :',ma,'the MSE is: ',simulation[ma]['final']['mse'])
print(text+'Experiment ',i+1,' for MA :',ma,'the RMSE is: ',simulation[ma]['final']['rmse'])
# print(text+'Experiment ',i+1,' for MA :',ma,'the MAE is: ',simulation[ma]['final']['mae'])
for ma in simulation.keys():
print(text+'Experiment ',i+1,' for MA :',ma,'the MSE is: ',simulation[ma]['final']['mse'])
# print(text+'Experiment ',i+1,' for MA :',ma,'the RMSE is: ',simulation[ma]['final']['rmse'])
# print(text+'Experiment ',i+1,' for MA :',ma,'the MAE is: ',simulation[ma]['final']['mae'])
for ma in simulation.keys():
# print(text+'Experiment ',i+1,' for MA :',ma,'the MSE is: ',simulation[ma]['final']['mse'])
# print(text+'Experiment ',i+1,' for MA :',ma,'the RMSE is: ',simulation[ma]['final']['rmse'])
print(text+'Experiment ',i+1,' for MA :',ma,'the MAE is: ',simulation[ma]['final']['mae'])
Stock with Sentiment Trends Experiment 1 for MA : SMA the RMSE is: 6.268680607174258 Stock with Sentiment Trends Experiment 1 for MA : EMA the RMSE is: 5.50005196725239 Stock with Sentiment Trends Experiment 1 for MA : WMA the RMSE is: 8.057694292448089 Stock with Sentiment Trends Experiment 1 for MA : DEMA the RMSE is: 7.279183760008 Stock with Sentiment Trends Experiment 1 for MA : KAMA the RMSE is: 5.796783486223054 Stock with Sentiment Trends Experiment 1 for MA : MIDPOINT the RMSE is: 4.768795618921396 Stock with Sentiment Trends Experiment 1 for MA : T3 the RMSE is: 8.434809584670473 Stock with Sentiment Trends Experiment 1 for MA : TEMA the RMSE is: 5.158323598407468 Stock with Sentiment Trends Experiment 1 for MA : SMA the MSE is: 39.29635655476263 Stock with Sentiment Trends Experiment 1 for MA : EMA the MSE is: 30.25057164247689 Stock with Sentiment Trends Experiment 1 for MA : WMA the MSE is: 64.92643731055051 Stock with Sentiment Trends Experiment 1 for MA : DEMA the MSE is: 52.98651621196421 Stock with Sentiment Trends Experiment 1 for MA : KAMA the MSE is: 33.6026987861483 Stock with Sentiment Trends Experiment 1 for MA : MIDPOINT the MSE is: 22.7414116550439 Stock with Sentiment Trends Experiment 1 for MA : T3 the MSE is: 71.14601272964887 Stock with Sentiment Trends Experiment 1 for MA : TEMA the MSE is: 26.608302345887367 Stock with Sentiment Trends Experiment 1 for MA : SMA the MAE is: 5.07416987111494 Stock with Sentiment Trends Experiment 1 for MA : EMA the MAE is: 4.444486270049439 Stock with Sentiment Trends Experiment 1 for MA : WMA the MAE is: 6.289003841800032 Stock with Sentiment Trends Experiment 1 for MA : DEMA the MAE is: 5.725540843661134 Stock with Sentiment Trends Experiment 1 for MA : KAMA the MAE is: 4.518981487962124 Stock with Sentiment Trends Experiment 1 for MA : MIDPOINT the MAE is: 3.944458615157319 Stock with Sentiment Trends Experiment 1 for MA : T3 the MAE is: 6.848574357624394 Stock with Sentiment Trends Experiment 1 for MA : TEMA the MAE is: 4.336839144940602 Stock with Sentiment Trends Experiment 2 for MA : SMA the RMSE is: 6.3146847095225125 Stock with Sentiment Trends Experiment 2 for MA : EMA the RMSE is: 8.584515095821342 Stock with Sentiment Trends Experiment 2 for MA : WMA the RMSE is: 8.76939754681257 Stock with Sentiment Trends Experiment 2 for MA : DEMA the RMSE is: 11.448153288374908 Stock with Sentiment Trends Experiment 2 for MA : KAMA the RMSE is: 6.49063413418937 Stock with Sentiment Trends Experiment 2 for MA : MIDPOINT the RMSE is: 10.135265684380567 Stock with Sentiment Trends Experiment 2 for MA : T3 the RMSE is: 10.333769318274479 Stock with Sentiment Trends Experiment 2 for MA : TEMA the RMSE is: 8.86953066007257 Stock with Sentiment Trends Experiment 2 for MA : SMA the MSE is: 39.87524298067742 Stock with Sentiment Trends Experiment 2 for MA : EMA the MSE is: 73.6938994303845 Stock with Sentiment Trends Experiment 2 for MA : WMA the MSE is: 76.90233333404232 Stock with Sentiment Trends Experiment 2 for MA : DEMA the MSE is: 131.0602137141292 Stock with Sentiment Trends Experiment 2 for MA : KAMA the MSE is: 42.128331463904196 Stock with Sentiment Trends Experiment 2 for MA : MIDPOINT the MSE is: 102.72361049298229 Stock with Sentiment Trends Experiment 2 for MA : T3 the MSE is: 106.78678832331099 Stock with Sentiment Trends Experiment 2 for MA : TEMA the MSE is: 78.66857412996737 Stock with Sentiment Trends Experiment 2 for MA : SMA the MAE is: 5.088204561858354 Stock with Sentiment Trends Experiment 2 for MA : EMA the MAE is: 7.207683534137507 Stock with Sentiment Trends Experiment 2 for MA : WMA the MAE is: 7.121360770950225 Stock with Sentiment Trends Experiment 2 for MA : DEMA the MAE is: 10.329401784343453 Stock with Sentiment Trends Experiment 2 for MA : KAMA the MAE is: 5.225311201416733 Stock with Sentiment Trends Experiment 2 for MA : MIDPOINT the MAE is: 8.372939983805384 Stock with Sentiment Trends Experiment 2 for MA : T3 the MAE is: 8.334873887974807 Stock with Sentiment Trends Experiment 2 for MA : TEMA the MAE is: 7.566655356961055 Stock with Sentiment Trends Experiment 3 for MA : SMA the RMSE is: 6.302148600548591 Stock with Sentiment Trends Experiment 3 for MA : EMA the RMSE is: 4.306552961339474 Stock with Sentiment Trends Experiment 3 for MA : WMA the RMSE is: 10.021183520932126 Stock with Sentiment Trends Experiment 3 for MA : DEMA the RMSE is: 6.536375949623308 Stock with Sentiment Trends Experiment 3 for MA : KAMA the RMSE is: 4.863998495275778 Stock with Sentiment Trends Experiment 3 for MA : MIDPOINT the RMSE is: 4.5973124715409295 Stock with Sentiment Trends Experiment 3 for MA : T3 the RMSE is: 10.09431101894578 Stock with Sentiment Trends Experiment 3 for MA : TEMA the RMSE is: 3.316477447623887 Stock with Sentiment Trends Experiment 3 for MA : SMA the MSE is: 39.71707698339657 Stock with Sentiment Trends Experiment 3 for MA : EMA the MSE is: 18.546398408821787 Stock with Sentiment Trends Experiment 3 for MA : WMA the MSE is: 100.42411916020163 Stock with Sentiment Trends Experiment 3 for MA : DEMA the MSE is: 42.724210554814 Stock with Sentiment Trends Experiment 3 for MA : KAMA the MSE is: 23.65848136204503 Stock with Sentiment Trends Experiment 3 for MA : MIDPOINT the MSE is: 21.13528196098577 Stock with Sentiment Trends Experiment 3 for MA : T3 the MSE is: 101.89511494721017 Stock with Sentiment Trends Experiment 3 for MA : TEMA the MSE is: 10.999022660597852 Stock with Sentiment Trends Experiment 3 for MA : SMA the MAE is: 5.205453851116191 Stock with Sentiment Trends Experiment 3 for MA : EMA the MAE is: 3.4160340524918316 Stock with Sentiment Trends Experiment 3 for MA : WMA the MAE is: 8.070563561893302 Stock with Sentiment Trends Experiment 3 for MA : DEMA the MAE is: 5.453197818871469 Stock with Sentiment Trends Experiment 3 for MA : KAMA the MAE is: 3.972687129543795 Stock with Sentiment Trends Experiment 3 for MA : MIDPOINT the MAE is: 3.8079919461917573 Stock with Sentiment Trends Experiment 3 for MA : T3 the MAE is: 8.111172771475383 Stock with Sentiment Trends Experiment 3 for MA : TEMA the MAE is: 2.655228978622383 Stock with Sentiment Trends Experiment 4 for MA : SMA the RMSE is: 5.008128583557969 Stock with Sentiment Trends Experiment 4 for MA : EMA the RMSE is: 6.153459823215273 Stock with Sentiment Trends Experiment 4 for MA : WMA the RMSE is: 7.306033595657539 Stock with Sentiment Trends Experiment 4 for MA : DEMA the RMSE is: 11.541090995320783 Stock with Sentiment Trends Experiment 4 for MA : KAMA the RMSE is: 4.549058713304123 Stock with Sentiment Trends Experiment 4 for MA : MIDPOINT the RMSE is: 4.271324788049093 Stock with Sentiment Trends Experiment 4 for MA : T3 the RMSE is: 8.655730725901707 Stock with Sentiment Trends Experiment 4 for MA : TEMA the RMSE is: 5.799675429452453 Stock with Sentiment Trends Experiment 4 for MA : SMA the MSE is: 25.081351909450348 Stock with Sentiment Trends Experiment 4 for MA : EMA the MSE is: 37.86506779592453 Stock with Sentiment Trends Experiment 4 for MA : WMA the MSE is: 53.37812690087662 Stock with Sentiment Trends Experiment 4 for MA : DEMA the MSE is: 133.19678136227444 Stock with Sentiment Trends Experiment 4 for MA : KAMA the MSE is: 20.693935177088164 Stock with Sentiment Trends Experiment 4 for MA : MIDPOINT the MSE is: 18.24421544500263 Stock with Sentiment Trends Experiment 4 for MA : T3 the MSE is: 74.9216743993189 Stock with Sentiment Trends Experiment 4 for MA : TEMA the MSE is: 33.63623508699449 Stock with Sentiment Trends Experiment 4 for MA : SMA the MAE is: 3.9377037384058786 Stock with Sentiment Trends Experiment 4 for MA : EMA the MAE is: 4.830217084369749 Stock with Sentiment Trends Experiment 4 for MA : WMA the MAE is: 5.95316326588041 Stock with Sentiment Trends Experiment 4 for MA : DEMA the MAE is: 10.29859546777107 Stock with Sentiment Trends Experiment 4 for MA : KAMA the MAE is: 3.6577262429810227 Stock with Sentiment Trends Experiment 4 for MA : MIDPOINT the MAE is: 3.3887721441386436 Stock with Sentiment Trends Experiment 4 for MA : T3 the MAE is: 7.03412901576443 Stock with Sentiment Trends Experiment 4 for MA : TEMA the MAE is: 5.291282498785388 Stock with Sentiment Trends Experiment 5 for MA : SMA the RMSE is: 4.785964622427749 Stock with Sentiment Trends Experiment 5 for MA : EMA the RMSE is: 4.932463328983385 Stock with Sentiment Trends Experiment 5 for MA : WMA the RMSE is: 5.170235936522176 Stock with Sentiment Trends Experiment 5 for MA : DEMA the RMSE is: 4.53382842274277 Stock with Sentiment Trends Experiment 5 for MA : KAMA the RMSE is: 6.054666393710646 Stock with Sentiment Trends Experiment 5 for MA : MIDPOINT the RMSE is: 6.797920894183608 Stock with Sentiment Trends Experiment 5 for MA : T3 the RMSE is: 6.278727107833918 Stock with Sentiment Trends Experiment 5 for MA : TEMA the RMSE is: 5.217204420122223 Stock with Sentiment Trends Experiment 5 for MA : SMA the MSE is: 22.905457367129987 Stock with Sentiment Trends Experiment 5 for MA : EMA the MSE is: 24.329194491765858 Stock with Sentiment Trends Experiment 5 for MA : WMA the MSE is: 26.731339639305343 Stock with Sentiment Trends Experiment 5 for MA : DEMA the MSE is: 20.555600166870196 Stock with Sentiment Trends Experiment 5 for MA : KAMA the MSE is: 36.658985139129086 Stock with Sentiment Trends Experiment 5 for MA : MIDPOINT the MSE is: 46.211728483578064 Stock with Sentiment Trends Experiment 5 for MA : T3 the MSE is: 39.422414094648474 Stock with Sentiment Trends Experiment 5 for MA : TEMA the MSE is: 27.219221961342864 Stock with Sentiment Trends Experiment 5 for MA : SMA the MAE is: 4.003866138329542 Stock with Sentiment Trends Experiment 5 for MA : EMA the MAE is: 4.099346553838579 Stock with Sentiment Trends Experiment 5 for MA : WMA the MAE is: 4.142288801040536 Stock with Sentiment Trends Experiment 5 for MA : DEMA the MAE is: 3.6522177332314283 Stock with Sentiment Trends Experiment 5 for MA : KAMA the MAE is: 4.91375972579294 Stock with Sentiment Trends Experiment 5 for MA : MIDPOINT the MAE is: 5.510818514624332 Stock with Sentiment Trends Experiment 5 for MA : T3 the MAE is: 5.177910469752962 Stock with Sentiment Trends Experiment 5 for MA : TEMA the MAE is: 4.028826161355838 Stock with Sentiment Trends Experiment 6 for MA : SMA the RMSE is: 10.442935103170173 Stock with Sentiment Trends Experiment 6 for MA : EMA the RMSE is: 8.546427546622475 Stock with Sentiment Trends Experiment 6 for MA : WMA the RMSE is: 9.21682096296829 Stock with Sentiment Trends Experiment 6 for MA : DEMA the RMSE is: 12.313030072070955 Stock with Sentiment Trends Experiment 6 for MA : KAMA the RMSE is: 8.944350180476702 Stock with Sentiment Trends Experiment 6 for MA : MIDPOINT the RMSE is: 7.822007405365507 Stock with Sentiment Trends Experiment 6 for MA : T3 the RMSE is: 10.514998001927937 Stock with Sentiment Trends Experiment 6 for MA : TEMA the RMSE is: 8.340715527022663 Stock with Sentiment Trends Experiment 6 for MA : SMA the MSE is: 109.05489356902385 Stock with Sentiment Trends Experiment 6 for MA : EMA the MSE is: 73.04142380966745 Stock with Sentiment Trends Experiment 6 for MA : WMA the MSE is: 84.94978866341171 Stock with Sentiment Trends Experiment 6 for MA : DEMA the MSE is: 151.61070955572364 Stock with Sentiment Trends Experiment 6 for MA : KAMA the MSE is: 80.0014001509936 Stock with Sentiment Trends Experiment 6 for MA : MIDPOINT the MSE is: 61.18379984959283 Stock with Sentiment Trends Experiment 6 for MA : T3 the MSE is: 110.56518298054853 Stock with Sentiment Trends Experiment 6 for MA : TEMA the MSE is: 69.56753550271695 Stock with Sentiment Trends Experiment 6 for MA : SMA the MAE is: 8.732625329283675 Stock with Sentiment Trends Experiment 6 for MA : EMA the MAE is: 7.099244401385842 Stock with Sentiment Trends Experiment 6 for MA : WMA the MAE is: 7.490547440692417 Stock with Sentiment Trends Experiment 6 for MA : DEMA the MAE is: 11.085595013418024 Stock with Sentiment Trends Experiment 6 for MA : KAMA the MAE is: 7.358601729961256 Stock with Sentiment Trends Experiment 6 for MA : MIDPOINT the MAE is: 6.441728946960992 Stock with Sentiment Trends Experiment 6 for MA : T3 the MAE is: 8.473394546481362 Stock with Sentiment Trends Experiment 6 for MA : TEMA the MAE is: 7.185876850367952 Stock with Sentiment Trends Experiment 7 for MA : SMA the RMSE is: 11.561505870388714 Stock with Sentiment Trends Experiment 7 for MA : EMA the RMSE is: 5.876165364129462 Stock with Sentiment Trends Experiment 7 for MA : WMA the RMSE is: 6.061739426637149 Stock with Sentiment Trends Experiment 7 for MA : DEMA the RMSE is: 9.351899538028238 Stock with Sentiment Trends Experiment 7 for MA : KAMA the RMSE is: 7.460294825752223 Stock with Sentiment Trends Experiment 7 for MA : MIDPOINT the RMSE is: 5.346491199686541 Stock with Sentiment Trends Experiment 7 for MA : T3 the RMSE is: 11.301752280128088 Stock with Sentiment Trends Experiment 7 for MA : TEMA the RMSE is: 5.8480564200717415 Stock with Sentiment Trends Experiment 7 for MA : SMA the MSE is: 133.6684179910327 Stock with Sentiment Trends Experiment 7 for MA : EMA the MSE is: 34.52931938659473 Stock with Sentiment Trends Experiment 7 for MA : WMA the MSE is: 36.74468487644727 Stock with Sentiment Trends Experiment 7 for MA : DEMA the MSE is: 87.45802496937279 Stock with Sentiment Trends Experiment 7 for MA : KAMA the MSE is: 55.655998887145394 Stock with Sentiment Trends Experiment 7 for MA : MIDPOINT the MSE is: 28.58496814832563 Stock with Sentiment Trends Experiment 7 for MA : T3 the MSE is: 127.72960460138046 Stock with Sentiment Trends Experiment 7 for MA : TEMA the MSE is: 34.19976389234231 Stock with Sentiment Trends Experiment 7 for MA : SMA the MAE is: 10.289389775089397 Stock with Sentiment Trends Experiment 7 for MA : EMA the MAE is: 4.852473639818076 Stock with Sentiment Trends Experiment 7 for MA : WMA the MAE is: 4.85767480758183 Stock with Sentiment Trends Experiment 7 for MA : DEMA the MAE is: 8.239361009856534 Stock with Sentiment Trends Experiment 7 for MA : KAMA the MAE is: 6.325008398714769 Stock with Sentiment Trends Experiment 7 for MA : MIDPOINT the MAE is: 4.412567377496203 Stock with Sentiment Trends Experiment 7 for MA : T3 the MAE is: 9.16046603172084 Stock with Sentiment Trends Experiment 7 for MA : TEMA the MAE is: 5.017065986818192 Stock with Sentiment Trends Experiment 8 for MA : SMA the RMSE is: 4.880975519218278 Stock with Sentiment Trends Experiment 8 for MA : EMA the RMSE is: 6.043133875644865 Stock with Sentiment Trends Experiment 8 for MA : WMA the RMSE is: 6.622855684214975 Stock with Sentiment Trends Experiment 8 for MA : DEMA the RMSE is: 11.91961743593652 Stock with Sentiment Trends Experiment 8 for MA : KAMA the RMSE is: 4.694565604625168 Stock with Sentiment Trends Experiment 8 for MA : MIDPOINT the RMSE is: 4.298904276636815 Stock with Sentiment Trends Experiment 8 for MA : T3 the RMSE is: 8.813596908808496 Stock with Sentiment Trends Experiment 8 for MA : TEMA the RMSE is: 5.656001741438255 Stock with Sentiment Trends Experiment 8 for MA : SMA the MSE is: 23.82392201920814 Stock with Sentiment Trends Experiment 8 for MA : EMA the MSE is: 36.51946703896653 Stock with Sentiment Trends Experiment 8 for MA : WMA the MSE is: 43.8622174139386 Stock with Sentiment Trends Experiment 8 for MA : DEMA the MSE is: 142.0772798190819 Stock with Sentiment Trends Experiment 8 for MA : KAMA the MSE is: 22.038946216129663 Stock with Sentiment Trends Experiment 8 for MA : MIDPOINT the MSE is: 18.4805779796863 Stock with Sentiment Trends Experiment 8 for MA : T3 the MSE is: 77.67949047095865 Stock with Sentiment Trends Experiment 8 for MA : TEMA the MSE is: 31.990355699152573 Stock with Sentiment Trends Experiment 8 for MA : SMA the MAE is: 3.848681389687446 Stock with Sentiment Trends Experiment 8 for MA : EMA the MAE is: 4.745071953502077 Stock with Sentiment Trends Experiment 8 for MA : WMA the MAE is: 5.251112001385055 Stock with Sentiment Trends Experiment 8 for MA : DEMA the MAE is: 10.62322903609799 Stock with Sentiment Trends Experiment 8 for MA : KAMA the MAE is: 3.7548184287573756 Stock with Sentiment Trends Experiment 8 for MA : MIDPOINT the MAE is: 3.422385888494848 Stock with Sentiment Trends Experiment 8 for MA : T3 the MAE is: 7.190279750268759 Stock with Sentiment Trends Experiment 8 for MA : TEMA the MAE is: 5.124316314221195
Create HTML¶
In [2]:
from google.colab import drive
drive.mount('/content/drive')
Mounted at /content/drive
In [115]:
cd ..
/content/drive/.shortcut-targets-by-id/1IaGjVBlTspxI2CHSrxfYnaiYvsaG0pHs/Stock price prediction/Archana - LSTM Hybrid/Outputs
In [5]:
cd drive/MyDrive/Stock price prediction/Archana - LSTM Hybrid
/content/drive/.shortcut-targets-by-id/1IaGjVBlTspxI2CHSrxfYnaiYvsaG0pHs/Stock price prediction/Archana - LSTM Hybrid
In [ ]:
%%shell
jupyter nbconvert --to html LSTM_Hybrid_using_TA_LIB_BullBear.ipynb